Structure/Function Analysis of Interleukin-2-Toxin (DABd8&L-2) FRAGMENT B SEQUENCES REQUIRED FOR THE DELIVERY OF FRAGMENT A TO THE CYTOSOL OF TARGET CELLS*

We used cassette and deletion mutagenesis to analyze the structural features of fragment B-related sequences in the fusion toxin DABbee-IL-2 (where IL-2 represents interleukin-2) that are necessary for the efficient delivery of fragment A to the cytosol of target cells. We demonstrate that whereas an intact disulfide between and Cys4” be required for the cytotoxic action of native this is not required for the cytotoxic of The in-frame deletion the dissociation

We have used cassette and deletion mutagenesis to analyze the structural features of fragment B-related sequences in the fusion toxin DABbee-IL-2 (where IL-2 represents interleukin-2) that are necessary for the efficient delivery of fragment A to the cytosol of target cells. We demonstrate that whereas an intact disulfide bond between Cys'*' and Cys4" may be required for the cytotoxic action of native diphtheria toxin, this bond is not required for the cytotoxic action of DABlse-IL-2. The in-frame deletion of the 97 amino acids from Thr"' to His"' of DAB,+se-IL-2 increases both the potency and the apparent dissociation constant (Kd) of the resulting fusion toxin for high affinity interleukin-2 receptor-bearing target cells. In contrast, the inframe deletion of either the 191 amino acids between AspZel and G1y483 or the 85 amino acids between AsnZo4 and IleZeo results in a lOOO-fold loss in potency. These regions contain the putative membrane-spanning regions and the amphipathic membrane surface-associating regions of fragment B, respectively. These results indicate that the efficient delivery of the ADPribosyltransferase from DAB4se-IL-2 to the cytosol requires the membrane-associating domains of fragment B. This function has been postulated to play a role in the diphtherial intoxication of eukaryotic cells. However, unlike native diphtheria toxin, fragment B sequences distal to Thr387 do not enhance the potency of DABAse-IL-2.
Recently, Williams et al. (1) have described the construction of a diphtheria toxin (DT)'-related fusion protein in which the native receptor-binding domain of DT was genetically replaced with the polypeptide hormone interleukin-2 (IL-2). IL-2-toxin (hereafter designated DAB4,,-IL-2) is a 68,202-Da fusion protein in which amino acids 2-133 of mature human IL-2 were joined in correct translational reading frame to Ala4= of DT fragment B. DAB,,-IL-Z has been shown to bind to the IL-2 receptor and to selectively intoxicate lymphocytes which bear the high affinity form of the IL-2 receptor (2, 3). Moreover, the cytotoxic action of DAB4s6-IL-2, like that of native diphtheria toxin, requires receptor-mediated endocytosis, passage through an acidic compartment, and delivery of fragment A-associated ADP-ribosyltransferase to the cytosol of target cells (2).
To further our understanding of the entry of DABdss-IL-2 fragment A into the cytosol and to define the role that fragment B sequences play in this process, we have used cassette exchange and in-frame deletion mutagenesis to assemble a family of DAB-IL-2 variants. We demonstrate that whereas a Cys471 to Tyr"" mutation renders DT nontoxic, the analogous mutation in DAB,,-IL-2 has little, if any, effect on the potency of this fusion toxin. Moreover, we show that fragment B sequences between Thr387 and His4@' can be deleted from DAB,,-IL-2 without a loss in biological activity, whereas the further deletion of Serzg2 to Thr387, a region which contains two putative membrane-spanning helices of fragment B, essentially abolishes cytotoxic activity. In a similar fashion, the deletion of fragment B sequences between Asnzo5 and IlezgO, which include an amphipathic region(s) predicted to have membrane surface binding properties, also results in a marked loss of cytotoxic activity.

AND DISCUSSION
The nontoxic diphtheria toxin-related protein CRMlOOl has been shown to result from a single G:A point mutation which changes Cys471 to Tyr471 (15). This mutation disrupts the disulfide bond between Cys461 and Cys471 and results in a loss of cytotoxic activity. Zucker (16) has shown by Schild analysis that CRMlOOl blocks the action of native toxin on Chinese hamster ovary cells as effectively as CRM197. Recently, Dell'Arciprete et al. (17) have obtained similar results for Vero cells. Since CRMlOOl retains ADP-ribosyltransferase activity and binds to receptors on Chinese hamster ovary cells (16) and Vero cells (17), its lack of cytotoxicity suggests that the fragment B disulfide bond may be required to retain a conformation necessary for efficient internalization and/or processing of receptor-bound native toxin. To determine whether an intact disulfide bond between cyLP* and CYSTS* was required for the biological activity of DAB4M-IL-2, a 587-base pair ClaI-S&I restriction fragment from the tox-1001 allele was subcloned into plasmid pDW24  (1001)486-IL-2) was purified from whole cell extracts by immunoaffinity chromatography and high pressure liquid chromatography.
In marked contrast to CRMlOOl, however, the fusion toxin which carries the Cys4" to Tyr472 mutation (DAB(1001)486-IL2) was found to be as toxic for HUT 102/6TG cells as DAB4=-IL-2 (data not shown). These results demonstrate that whereas the fragment B disulfide bond between Cys461 and Cys471 may be required for the biological activity of native diphtheria toxin, this disulfide bond is not required for biological activity of the fusion toxin.
In addition to CRMlOOl, other nontoxic DT-related proteins have been used to study the structural requirements for the delivery of DT fragment A to the cytosol of target cells. For example, CRM45 (which lacks the diphtheria toxin receptor-binding domain but retains the membrane-associating domains of fragment B) (18-20) and CRM26 (which terminates upstream of these hydrophobic domains) have both been used by Bacha et al. (21) to form conjugate toxins. In this study, DT-related CRM26 and CRM45 were chemically conjugated to thyrotropin-releasing hormone (TRH). Whereas both CRM26-TRH and CRM45-TRH were found to bind to the TRH receptor and to retain ADP-ribosyltransferase activity, only CRM45-TRH was found to be cytotoxic for TRH receptor-bearing cells. These observations suggested that the membrane-associating domains of fragment B were necessary for the delivery of fragment A to the cytosol of target cells.
Since the disulfide bond between Cys4@ and Cys4" was not required for the cytotoxic action of DABhs6-IL-2, it was of interest to determine the DT fragment B sequences which were essential for the delivery of fragment A from the fusion toxin to the cytosol of target cells. Several in-frame deletion mutations were introduced into the fragment B encoding portion of the DAB,,G-IL-2 toxin gene (Fig. 2). The first mutant (DABas9-IL-2) was constructed by removing a 309base pair HpaII-SphI restriction fragment from pDW24 and replacing it with the oligonucleotide linker 261/274 (Table 2) to generate plasmid pDW27 (Fig. 1). This linker restores fragment B sequences from Pro383 to Thr387 and allows for inframe fusion to IL-2 sequences at this position.
Thus, in DAB389-IL-2, the 97 amino acids between Thr387 and His48" have been deleted, and the DT-related component of the fusion toxin is equivalent to CRM45.
In a similar fashion, a 191-amino acid in-frame deletion was constructed by removing a ClaI-SphI restriction fragment from pDW24 and replacing it with the oligonucleotide linker 292/293 (Table 2) to form plasmid pDW28, which encodes DABZs,-IL-2 ( Fig. 1). In this case, the in-frame deletion encompasses the putative membrane-spanning helices that have been predicted to play a role in the delivery of fragment A to the eukaryotic cell cytosol (19,20).
In marked contrast, the IC& of DABZ9,-IL-2 was -lOOO-fold greater (-4 X 1O-7 M). These results strongly suggest that fragment B sequences between Se?" and Thr387 are essential for the efficient delivery of fragment A, whereas sequences between Thr387 and Hi&'" are not.
Since it was possible that the low potency of DABZg5-IL-2 for HUT 102/6TG cells was related to altered binding to the high affinity IL-2 receptor, a series of competitive displacement experiments using lz51-labeled recombinant IL-2 were conducted. As shown in Fig. 5, both DAB389-IL-2 and DAB295-IL-2 were found to have an apparent Kd -3 times lower than that of DAB,,-IL-2 (& = 8 X lo-' uersus 2.5 X lo-' M).
However, it should be noted that whereas DABzgs-IL-2 binds more avidly to the high affinity IL-2 receptor than does DABbm-IL-2, its cytotoxic activity is at least lOOO-fold lower (Fig. 4). These results demonstrate that avid binding to the target receptor is not, in itself, sufficient for the biological activity of the DT-related IL-2 fusion toxins and that fragment B sequences between Se?" and Thr387 are essential for a post-receptor binding event in the intoxication process. It is of particular interest to note that fragment B sequences from Ser*'* to Thr387 contain the putative membrane-spanning helices that have been postulated to facilitate the membrane translocation of fragment A to the cytosol(19,20  Papini et al. (23) have proposed that the first step in the diphtherial intoxication process involves the interaction of the toxin with the surface of the eukaryotic cell membrane. Lambotte et al. (20) have identified an amphipathic region of fragment B which is homologous to the surface lipid-associating domain of apolipoprotein A-I. Therefore, it was postulated that this amphipathic region (amino acids 210-252 in DAB4%-IL-2) plays a role in the intoxication process. To test this hypothesis, the 85-amino acid encoding region from NsiI to C.!uI was deleted from both pDW24 and pDW27 to form plasmids pDW30 and pDW31, respectively (Figs. 1 and 2 and Table 2). Plasmids pDW30 and pDW31 encode DAB(A205-289),+a6-IL-2 and DAB(A205-289)38g-IL-2, respectively. Following ligation and transformation, the DAB-IL-2-related fusion proteins were expressed and purified as described above (Fig. 3). As shown in Fig. 4, the deletion of Leuzo5 to Val*" (which includes the amphipathic domain(s) of fragment B) results in an -lOOO-fold loss of cytotoxic activity against high affinity IL-2 receptor-positive cells in uitro. In this instance, however, the deletion reduced the ability of the fusion toxin to compete for binding to the high affinity IL-2 receptor. For example, DABda6-IL-2 was found to have a Kd of 2.5 X lo-' M, whereas DAB(A205-289)4ss-IL-2 was found to have a Kd of 1 X 10m7 M. Comparison of the apparent dissociation constants of the DAB-IL-2 fusion toxins suggests that this region of fragment B may stabilize the interaction between the fusion toxin and the cell surface. Recent studies by Cabiaux et al. (24) of the secondary structure of DT interacting with acid liposomes support such a model. Using polarized infrared spectroscopy, these investigators found that the amphipathic helical domain(s) of DT associated with the liposome surface at low pH such that the long axis of the helix was oriented perpendicular to the lipid acyl chains. Furthermore, the putative transmembrane domains of fragment B were found to associate parallel to the acyl chains of the liposomes. Although these experi-ments lend credence to the binding and cytotoxicity data obtained for the four DAB-IL-2 deletion mutants reported here, the deletions are large, and the results must be interpreted with caution. Site-directed mutagenesis experiments are currently underway to further define the structural features of both the amphipathic and membrane-spanning regions of DAB-IL-2 fragment B that are required for the delivery of fragment A to the cytosol of target cells.
Recently, Edwards et al. (25) have examined the receptor binding properties of a fusion toxin composed of transforming growth factor LY (TGFa) and a truncated version of Pseudomonas exotoxin A (PE40). Since the orientation of the functional domains of exotoxin A are opposite that of DT, receptor-binding domain substitution occurs at the N terminus. TGFLu-PE40 was found to bind to the epidermal growth factor receptor less effectively than native TGFcu. Interestingly, the analysis of internal in-frame deletion mutants of TGFa-PE40 demonstrated that the removal of the N-terminal 59 or 130 amino acids from the PE40 toxophore resulted in fusion proteins that were increasingly effective in their epidermal growth factor receptor binding properties. However, the truncated TGFa-PE fusion toxins were less cytotoxic than fulllength TGFa-PE40.
In In contrast to the results of Edwards et al. (25) and Colombatti et al. (26), the in-frame deletion of 97 amino acids from the fragment B domain of DABha6-IL-2 resulted in a fusion toxin that bound more effectively to the high affinity IL-2 receptor while retaining full cytotoxic potency for target cells. Although it is possible that IL-2-related sequences in DAB389-IL-2 partially compensate for the deleted fragment B sequences, the similarity in the dose-response curves of DAB4e6-IL-2 and DAB3s9-IL-2 suggests that any role sequences from Thr387 to His*= may play in the entry process is below the level of detection in the HUT 102/6TG assay system.
In conclusion, studies reported here further define the interaction between DAB-IL-2 fusion toxins and target lymphocytes. The removal of 97 amino acids between Thr387 and His485 alters the DAB-IL-2 conformation such that the IL-2 component of the fusion toxin binds to its receptor with higher affinity. Conversely, deletion of the amphipathic domain(s) contained within the Leuzo5 to Va12sg region of fragment B results in a 4-fold loss of receptor binding affinity. This observation suggests that this region of fragment B may interact with the membrane surface and stabilize the interaction between the DAB-IL-2 fusion toxins and the cell. However, when the amphipathic domain(s) is deleted from DABzag-IL-2, the resulting fusion toxin binds to receptors as avidly as DAB486-IL-2, but is at least lOOO-fold less cytotoxic.
Thus, it appears that this membrane surface interaction may not only stabilize binding to the IL-2 receptor, but may be required to juxtapose the membrane-spanning domains of fragment B with the endocytic vesicle membrane for the efficient delivery of fragment A to the cytosol. Finally, deletion of the putative membrane-spanning helices from DABda6-IL-2 increases binding affinity, but abolishes cytotoxic activity.