Primary structure of diphtheria toxin fragment B: structural similarities with lipid-binding domains

Mutagenic Deimmunization of Diphtheria Toxin for Use in Biologic Drug Development

BACKGROUND

Targeted toxins require multiple treatments and therefore must be deimmunized. We report a method of protein deimmunization based on the point mutation of highly hydrophilic R, K, D, E, and Q amino acids on the molecular surface of truncated diphtheria-toxin (DT390).

METHODS

Based on their surface position derived from an X-ray-crystallographic model, residues were chosen for point mutation that were located in prominent positions on the molecular surface and away from the catalytic site. Mice were immunized with a targeted toxin containing either a mutated DT390 containing seven critical point mutations or the non-mutated parental toxin form.

RESULTS

Serum analysis revealed a significant 90% reduction in anti-toxin antibodies in mice immunized with the mutant, but not the parental drug form despite multiple immunizations. The experiment was repeated in a second strain of mice with a different MHC-haplotype to address whether point mutation removed T or B cell epitopes. Findings were identical indicating that B cell epitopes were eliminated from DT. The mutant drug form lost only minimal activity in vitro as well as in vivo.

CONCLUSION

These findings indicate that this method may be effective for deimmunizing of other proteins and that discovery of a deimmunized form of DT may lead to the development of more effective targeted toxin.

Targeting myeloid leukemia with a DT(390)-mIL-3 fusion immunotoxin: ex vivo and in vivo studies in mice

The IL-3 receptor was expressed on a high frequency of myeloid leukemia cells and also on hematopoietic and vascular cells. We previously showed that a recombinant IL-3 fusion immunotoxin (DT(390)IL-3) expressed by splicing the murine IL-3 gene to a truncated diphtheria toxin (DT(390)) gene selectively killed IL-3R(+) expressing cells and was not uniformly toxic to uncommitted BM progenitor cells (Chan,C.-H., Blazar,B.R., Greenfield,L., Kreitman,R.J. and Vallera,D.A., 1996, Blood, 88, 1445-1456). Thus, we explored the feasibility of using DT(390)IL-3 as an anti-leukemia agent. DT(390)IL-3 was toxic when administered to mice at doses as low as 0.1 microg/day. The dose limiting toxicity appeared to be related to platelet and bleeding effects of the fusion toxin. Because of these effects, DT(390)IL-3 was studied ex vivo as a means of purging contaminating leukemia cells from BM grafts in a murine autologous BM transplantation. In this setting, as few as 1000 IL-3R-expressing, bcr/abl transformed myeloid 32Dp210 leukemia cells were lethal. An optimal purging interval of 10 nM/l for 8 h eliminated leukemia cells from 32Dp210/BM mixtures given to lethally irradiated (8 Gy) C3H/HeJ syngeneic mice. Mice given treated grafts containing BM and a lethal dose of 32Dp210 cells survived over 100 days while mice given untreated grafts did not survive (P < 0.00001). DT(390)IL-3 may prove highly useful for ex vivo purging of lethal malignant leukemia cells from autologous BM grafts.

Elimination of the disulphide bridge in fragment B of diphtheria toxin: effect on membrane insertion, channel formation, and ATP binding

Active diphtheria toxin consists of two disulphide-linked fragments, termed A and B. Fragment B, which contains an internal disulphide bridge, facilitates translocation of the enzymatically active fragment A to the cytosol of eukaryotic cells. In this process cation-selective channels are formed. An in vitro translated full-length mutant lacking the internal disulphide bridge (A-58**) was functionally indistinguishable from its disulphide-containing counterpart (A-58) with respect to trypsin sensitivity, receptor binding, A-fragment translocation, and channel formation. In contrast, the B fragment of A-58** (B-36**) was slightly less trypsin resistant than the S-S-containing B fragment, B-36, and was approximately 300-fold less efficient than B-36 in permeabilizing cells. When first dialysed and then reconstituted with A fragment, B fragment without disulphide bridge yielded a less-active toxin than did wild-type B fragment. We conclude that the disulphide bridge in fragment B is not necessary for toxicity, as earlier believed, and that channel formation may play a role in membrane translocation.

Ligand-dependent quenching of tryptophan fluorescence in human erythrocyte hexose transport protein

D-Glucose transport by the 492-residue human erythrocyte hexose transport protein may involve ligand-mediated conformational/positional changes. To examine this possibility, hydrophilic quencher molecules [potassium iodide and acrylamide (ACR)] were used to monitor the quenching of the total protein intrinsic fluorescence exhibited by the six protein tryptophan (Trp) residues in the presence and absence of substrate D-glucose, and in the presence of the inhibitors maltose and cytochalasin B. Protein fluorescence was found to be quenched under various conditions, ca. 14-24% by KI and ca. 25-33% by ACR, indicating that the bulk of the Trp residue population occurs in normally inaccessible hydrophobic regions of the erythrocyte membrane. However, in the presence of D-glucose, quenching by KI and ACR decreased an average of -3.4% and -4.4%, respectively; Stern-Volmer plots displayed decreased slopes in the presence of D-glucose, confirming the relatively reduced quenching. In contrast, quenching efficiency increased in the presence of maltose (+5.9%, +3.3%), while addition of cytochalasin B had no effect on fluorescence quenching. The overall results are interpreted in terms of ligand-activated movement of an initially aqueous-located protein segment containing a Trp residue into, or toward, the cellular membrane. Relocation of this segment, in effect, opens the D-glucose channel; maltose and cytochalasin B would thus inhibit transport by mechanisms which block this positional change. Conformational and hydropathy analyses suggested that the region surrounding Trp-388 is an optimal "dynamic segment" which, in response to ligand activation, could undergo the experimentally deduced aqueous/membrane domain transfer.

Sequence similarity between protein B and human apolipoprotein A-IV

Sequence comparison of protein B (CAMP-factor) with human apolipoprotein A-IV (apo A-IV) revealed 32% similarity between the N-terminal part of protein B and a part of the putative lipid-binding domain of apo A-IV. The significance of this similarity is discussed with respect to the structure/function relationship of protein B.

Expression of a biologically active diphtheria toxin fragment B in Escherichia coli

The toxB gene of Corynebacterium diphtheriae bacteriophage beta encoding the B fragment of diphtheria toxin was cloned into an inducible expression vector. When expressed in Escherichia coli, fragment B was not proteolysed and was indistinguishable, by immunological criteria, from wild-type C. diphtheriae-derived fragment B. Soluble fragment B was partially purified from the cytoplasm by saline precipitation steps and was shown to compete with the wild-type diphtheria toxin for binding to receptors of sensitive eukaryotic cells. A complete diphtheria toxin was reconstituted by formation of the disulphide bridge between purified fragment A and recombinant fragment B, which migrates at the expected Mr on Western blots and which was able to block protein synthesis by ADP-ribosylation of elongation factor-2, thereby indicating that the recombinant fragment B had retained its biological activity.

Common features of Bacillus thuringiensis toxins specific for Diptera and Lepidoptera

The complete nucleotide sequence of a cloned gene encoding a 130-kDa crystal protein of Bacillus thuringiensis (B.t.) subspecies israelensis has been determined. The recombinant protein (Bt8) was purified and shown to be a mosquito-specific toxin with a LC50 value of 43 ng/ml to third-instar larvae of Aedes aegypti. Bt8 is processed by proteases or midgut extracts of mosquito larvae into toxic fragments of 68-78 kDa. Deletion mapping indicated that the active fragment of Bt8 is localized in the N-terminal half of the protoxin molecule. The deduced amino acid sequence of Bt8 has been compared with that of Bt2, a Lepidoptera-specific toxin, previously cloned from Bacillus thuringiensis berliner. Highly homologous amino acid stretches are present in the C-terminal half of the proteins. The N-terminal parts show much less sequence homology but they display a strikingly similar distribution of hydrophilic and hydrophobic amino acids. In addition, Bt8 and Bt2 show a significant immunological cross-reaction. The data indicate that although these B.t. delta endotoxins exhibit a different insect-host specificity, they are structurally related and might use a similar mechanism to interact with insect cell membranes.

Interleukin 2 toxin: a step toward selective immunomodulation

We have used protein engineering and recombinant DNA methodologies to genetically replace the eukaryotic cell receptor binding domain of diphtheria toxin with interleukin 2 (IL-2). The toxin-related T cell growth factor fusion gene has been cloned in Escherichia coli K12. Recombinant strains of E coli produce a 68,086 K hybrid toxin, IL-2 toxin that retains immunologic properties intrinsic to both its diphtheria toxin and IL-2 components. IL-2 toxin has been found to selectively inhibit protein synthesis in both human and murine T cell lines that bear high affinity IL-2 receptors, whereas the hybrid toxin is not active against cells that do not bear this receptor. The cytotoxic action of IL-2 toxin is specifically blocked by free IL-2 and monoclonal antibodies that bind to the p55 (Tac antigen) subunit of the high affinity IL-2 receptor. In addition, IL-2 toxin, like diphtheria toxin itself, must pass through an acidic compartment in order to deliver its adenosine diphosphate ribosyl transferase activity to the cytosol of target T cells. In a murine delayed type hypersensitivity (DTH) model system, we have shown that IL-2 toxin treatment induces a marked immunosuppression.

Lipid interaction of diphtheria toxin and mutants with altered fragment B. 1. Liposome aggregation and fusion

The interaction of diphtheria toxin and its cross-reacting mutants crm 45,228 and 1001 with small unilamellar vesicles has been followed by a turbidity assay, electron microscopy, fluorescence energy transfer and membrane permeability. All toxins at pH lower than 6 induce the aggregation and fusion of liposomes containing negatively charged phospholipids; crm 45 and crm 1001 are less potent than diphtheria toxin. Isolated diphtheria toxin fragment B is very effective while isolated fragment A is ineffective. Liposome fusion induced by the toxins at low pH occurs without release of the internal content implying that fusion does not involve vesicle breakage and resealing. The pH dependence of the membrane interaction of diphtheria toxin monitored by turbidity is in close agreement with that monitored by fluorescence energy transfer. It shows that diphtheria toxin can alter the lipid bilayer structure in the pH interval 5-6. This pH range occurs in endosomes and suggests that histidyl and carboxyl residues are likely to be involved in the conformational change of diphtheria toxin triggered by acidic pH.

Cloning and expression in Escherichia coli of three fragments of diphtheria toxin truncated within fragment B

We have constructed three different truncated versions of diphtheria toxin (a 535-amino-acid polypeptide) which correspond to the N-terminal 290, 377, and 485 amino acids of the toxin. These lengths include one, three, and all four of the putative membrane-spanning sequences of the toxin which are thought to play a role in the translocation of fragment A into cells. Each of these three genes has been modified at its 3' end to code for a C-terminal cysteine (to allow for disulfide linkage of a targeting ligand) or a gene fusion with alpha-melanocyte-stimulating hormone. We have also substituted the native diphtheria tox promoter (ptox) with the lambda pR promoter in an effort to overexpress these proteins. The truncated genes are expressed in Escherichia coli from both the tox promoter in a constitutive fashion and from the pR promoter by using the heat-inducible cI857 repressor. The clones produce proteins which react with anti-diphtheria toxin serum, which migrate at the anticipated Mr on Western blots, and which have ADP-ribosyltransferase activity. Constitutive synthesis from ptox leads to severe proteolytic degradation even in a protease-deficient strain. High-level expression from the pR promoter in the same lon htpR strain allows the full-length polypeptides to accumulate but also stops the growth of the cells. It appears that removal of as few as 50 amino acids from the C-terminus of diphtheria toxin alters its conformation, making it a target for proteases and causing overexpression lethality in the host cells.

Dimeric form of diphtheria toxin: purification and characterization

Many preparations of diphtheria toxin were found to contain dimeric and multimeric toxin forms. The monomeric and dimeric forms were fractionated to greater than 98% purity, and their properties were compared. Dimeric toxin slowly dissociated to native monomers in solution at neutral pH and could be rapidly dissociated with dimethyl sulfoxide. In cell culture assays and rabbit skin tests, the dimer exhibited no significant toxic activity, except for that attributable to trace contamination by monomer, or partial dissociation to monomer during the incubation period. In guinea pig lethality tests, however, toxic activity varied depending upon the dose. At least 7-fold greater amounts of dimer than monomer (161 ng vs. 22 ng, respectively) were required to cause death at 18 h, whereas similar weights of the two toxin forms (22 ng) caused death at 120 h. This variability probably reflected slow dissociation of dimer to monomer in the animal. The dimer was unable to bind toxin receptors on the surface of susceptible cells, whereas it retained full activity in the ADP-ribosyltransferase, NAD-glycohydrolase, or ligand-binding assays. Thus, the lack of toxicity of the dimeric toxin may have resulted from distortion or occlusion of the receptor binding site on the B moiety. We propose that the dimer contains two monomeric units bound by hydrophobic interactions and that the points of contact involve regions of the B moieties that are normally buried in the native monomer.

Structure of exotoxin A of Pseudomonas aeruginosa at 3.0-Angstrom resolution

Exotoxin A of Pseudomonas aeruginosa is a secreted bacterial toxin capable of translocating a catalytic domain into mammalian cells and inhibiting protein synthesis by the ADP-ribosylation of cellular elongation factor 2. The protein is a single polypeptide chain of 613 amino acids. The x-ray crystallographic structure of exotoxin A, determined to 3.0-A resolution, shows the following: an amino-terminal domain, composed primarily of antiparallel beta-structure and comprising approximately half of the molecule; a middle domain composed of alpha-helices; and a carboxyl-terminal domain comprising approximately one-third of the molecule. The carboxyl-terminal domain is the ADP-ribosyltransferase of the toxin. The other two domains are presumably involved in cell receptor binding and membrane translocation.

Evidence that membrane phospholipids and protein are required for binding of diphtheria toxin in Vero cells

Treatment with phospholipase C strongly protected monkey kidney (Vero) cells against diphtheria toxin and reduced the ability of the cells to bind 125I-labelled toxin. Treatment with phospholipase D and with trypsin also protected the cells, although to a lesser extent. Phospholipase A2 had no protective effect. Phospholipase C also protected fetal hamster kidney cells against the toxin. After removal of the enzymes, as well as after treatment of the cells with 4-acetamide 4'-isothiocyanostilbene 2,2'-disulfonic acid, diphtheria toxin binding capability was restored slowly, apparently by a process requiring protein synthesis, since cycloheximide blocked the restoration. The data indicate that both phospholipids and protein are involved in the binding sites for diphtheria toxin.

Tetanus toxin induces fusion and aggregation of lipid vesicles containing phosphatidylinositol at low pH

We report here on the ability of tetanus toxin to induce, at low pH, fusion and aggregation of lipid vesicles containing phosphatidylinositol. It has been shown that diphtheria toxin is internalized in acidic vacuoles (endosomes) and that the low endosomal pH could induce a protein conformational change responsible for the interaction with the endosomal membranes and the toxin translocation into the cytoplasm. The data here reported indicate that tetanus toxin might interact with lipid membrane in a similar way as diphtheria toxin suggesting for the two proteins an identical mechanism of entry into cells.

Aggregation and fusion of lipid vesicles induced by diphtheria toxin at low pH: possible involvement of the P site and the NAD+ binding site

Model membranes have been used to study the interaction between diphtheria toxin and lipids. We report here on the ability of this toxin to induce, at low pH, fusion and aggregation of asolectin lipid vesicles. Resonance energy transfer experiments using lipid fluorescent probes make it possible to discriminate between these two processes.

The complete amino acid sequence of diphtheria toxin fragment B. Correlation with its lipid-binding properties

The complete amino acid sequence of fragment B from diphtheria toxin has been determined. The polypeptide chain was split with cyanogen bromide, o-iodosobenzoic acid, clostripain and trypsin; all amino acid sequence analyses were made by automated Edman degradation. Fragment B, which corresponds to the carboxy terminus of the toxin molecule, contains 342 amino acids and has an Mr of 37240. The proposed amino acid sequence fully confirms the structure recently deduced from the nucleotide sequence of the structural gene. The complete sequence is analyzed in relationship with the role of fragment B in the transfer of diphtheria toxin fragment A from the extracellular medium into the cell cytoplasm.

Structure and action of heteronemertine polypeptide toxins: importance of amphipathic helix for activity of Cerebratulus lacteus toxin A-III

The marine heteronemertine Cerebratulus lacteus produces a family of protein cytolysins designated as A-toxins. Limited subtilisin digests of the most abundant homolog, toxin A-III, yield two major products which may be purified by high-performance liquid chromatography. One product is shown to represent residues 1-86 and the other contains the entire toxin sequence (1-95). Both polypeptides are shown to lack internal protease nicks. The 1-95 polypeptide retains full cytolytic activity in comparison to native toxin, whereas 1-86 has an activity that is approximately four times lower. Extensive treatment of A-III with carboxypeptidase Y yields a polypeptide containing residues 1-75 which is totally devoid of hemolytic activity. Residues 63-95 of native A-III have been predicted to form a relatively hydrophobic alpha-helix which is potentially important for activity. The circular dichroism spectrum of 1-95 is in excellent agreement with both experimental and Chou-Fasman-predicted secondary structures of native A-III, while the spectra of 1-86 and 1-75 indicate a loss of helicity quantitatively consistent with the removal of residues 87-95 and 76-95, respectively. Combined with our earlier data on bilayer penetration by N-terminal sequences (K. M. Blumenthal (1982) Biochemistry 21, 4229-4233], the current results indicate a direct involvement of both ends of A-III in lytic activity. The C-terminal region may function by contributing a membrane binding site in the form of an amphipathic helix.

Diphtheria toxin induces fusion of small unilamellar vesicles at low pH

Model membrane systems such as phospholipid vesicles have been extensively used to study the mechanism of membrane fusion at the molecular level. We report here on the capacity of diphtheria toxin to induce fusion of small unilamellar vesicles of dipalmitoylphosphatidylcholine at low pH. Fluorescence polarization and differential scanning calorimetry make it possible to demonstrate the mixing of the lipid phase. Mixing of the internal aqueous compartments of liposome was established using the terbium fluorescence technique. The analogy of structure and properties between melittin and a diphtheria toxin fragment is discussed.

Nucleotide sequence of the structural gene for diphtheria toxin carried by corynebacteriophage beta

A 1,942-base-pair DNA segment encoding the structural gene for diphtheria toxin was sequenced, and the primary structure of the toxin was deduced. Restriction enzyme fragments corresponding to nontoxic or hypotoxic peptides of the toxin were isolated from corynebacteriophage beta and cloned into Escherichia coli on plasmid pBR322, and the sequence was determined. The mature toxin molecule deduced from the sequence has 535 amino acid residues and a molecular weight of 58,342. The deduced sequence for the fragment A moiety was the same as that determined at the protein level, except for a single serine residue, which had been mispositioned in the earlier study. Several differences were noted with respect to the partial sequence data available on the fragment B moiety, some or all of which may reflect genetic variations among populations of corynephages carrying the toxin gene. The DNA sequence predicts a 25-residue leader peptide preceding the mature protein, which is presumably involved in secretion of the toxin from lysogenized Corynebacterium diphtheriae. We infer that initiation of translation probably occurs at a GTG codon (codon -25). Cloned restriction fragments containing sequences for the amino-terminal region of toxin, together with 5' flanking regions, were expressed in E. coli. Toxin-related peptides were synthesized and secreted into the periplasmic space. These results provide a basis for applying recombinant DNA methods to the study of diphtheria toxin and for producing novel, genetically altered forms of the toxin suited to the construction of new classes of immunotoxins.

Localization in diphtheria toxin fragment B of a region that induces pore formation in planar lipid bilayers at low pH

Like diphtheria toxin and the N-terminal (Mr 23 000) region of fragment B, CB1 (Mr 13 000), the cyanogen bromide peptide located in the middle region of fragment B is able to induce pore formation in lipid bilayer membrane at low pH. These two peptides (Mr 23 000 and 13 000) share a common segment (Mr 6300) containing the predicted amphipathic, alpha-helical, transverse lipid-associating domain (Mr 2750) of fragment B [J. Cell Biol. (1980) 87, 837-840]. Therefore, we postulated this domain to be responsible for the pore formation ability of diphtheria toxin [Proc. Natl. Acad. Sci. USA (1981) 78, 172-176]. A relationship between the pH dependency of pore formation and the presence of a cluster of prolines in the C-terminal region of CB1 is proposed.

Gating of ion channels made by a diphtheria toxin fragment in phospholipid bilayer membranes

B45, a fragment containing the major hydrophobic region of diphtheria toxin, increases the conductance of thin lipid membranes by forming ion-conducting channels that are gated by transmembrane voltage, Vm, and the bath pH. Single-channel currents show "bursting" behavior in the form of rapid transitions between a closed and an open conductance level. The average duration of a current "burst," as well as the total time a channel is actually open within a burst, decreases with increasing Vm. Analysis of these data suggests that, over a range of Vm, increases in the rate constants for transitions from the open to the closed states largely account for the decline in macroscopic conductance with increasing Vm. Increases in rate constants for transitions from a closed to an open conductance state are more likely to account for the increase in macroscopic conductance with increasing bath pH. Since several diphtheria toxin fragments and mutants are currently available, each containing various portions of the B45 region, it may be possible to study the relationship of the structure of these complex proteins to the detailed gating properties of the ion channels that they form.

Characteristics of guinea-pig immune sera elicited by a synthetic diphtheria toxin oligopeptide

Several oligopeptides of different lengths contained within the Cys 186-Cys 201 first disulfide loop of the diphtheria toxin molecule have been synthesized by a solid-phase method. 125I-labeled rabbit antibodies raised against diphtheria toxin reacted specifically with oligopeptides linked to m-nitrobenzhydrylamine resin when the amino acid chain length was equal to or greater than 10 residues. The synthetic tetradecapeptide (STDP) corresponding to the sequence Gly 188-Cys 201 was used to immunize guinea-pigs. The immune sera obtained reacted with the whole diphtheria toxin molecule as judged by an antigen-linked immunosorbent assay. Anti-STDP sera exhibited a clear, albeit limited, neutralizing effect against the lethal action of diphtheria toxin on cultivated Vero cells. The anti-STDP sera were also able to partially block the ADP-ribosylation of elongation factor 2 mediated by whole diphtheria toxin. In contrast, anti-STDP sera were almost inactive on the enzymic activities of either toxin fragment A or crm 45, a mutant protein which lacks the 15,000 mol. wt C-terminal sequence of the toxin molecule. On the basis of the results obtained, a possible localization of the Cys 188-Cys 201 loop region on the toxin molecule is proposed.

Structural prediction of membrane-bound proteins

A prediction algorithm based on physical characteristics of the twenty amino acids and refined by comparison to the proposed bacteriorhodopsin structure was devised to delineate likely membrane-buried regions in the primary sequences of proteins known to interact with the lipid bilayer. Application of the method to the sequence of the carboxyl terminal one-third of bovine rhodopsin predicted a membrane-buried helical hairpin structure. With the use of lipid-buried segments in bacteriorhodopsin as well as regions predicted by the algorithm in other membrane-bound proteins, a hierarchical ranking of the twenty amino acids in their preferences to be in lipid contact was calculated. A helical wheel analysis of the predicted regions suggests which helical faces are within the protein interior and which are in contact with the lipid bilayer.

Role of membrane gangliosides in the binding and action of bacterial toxins

Gangliosides are complex glycosphingolipids that contain from one to several residues of sialic acid. They are present in the plasma membrane of vertebrate cells with their oligosaccharide chains exposed to the external environment. They have been implicated as cell surface receptors and several bacterial toxins have been shown to interact with them. Cholera toxin, which mediates its effects on cells by activating adenylate cyclase, bind with high affinity and specificity to ganglioside GM1. Toxin-resistant cells which lack GM1 can be sensitized to cholera toxin by treating them with GM1. Cholera toxin specifically protects GM1 from cell surface labeling procedures and only GM1 is recovered when toxin-receptor complexes are isolated by immunoadsorption. These results clearly demonstrate that GM1 is the specific and only receptor for cholera toxin. Although cholera toxin binds to GM1 on the external side of the plasma membrane, it activates adenylate cyclase on the cytoplasmic side of the membrane by ADP-ribosylation of the regulatory component of the cyclase. GM1 in addition to functioning as a binding site for the toxin appears to facilitate its transmembrane movement. The heat-labile enterotoxin of E. coli is very similar to cholera toxin in both form and function and can also use GM1 as a cell surface receptor. The potent neurotoxin, tetanus toxin, has a high affinity for gangliosides GD1b and GT1b and binds to neurons which contain these gangliosides. It is not yet clear whether these gangliosides are the physiological receptors for tetanus toxin. By applying the techniques that established GM1 as the receptor for cholera toxin, the role of gangliosides as receptors for tetanus toxin as well as physiological effectors may be elucidated.