HB-EGF inhibitor derived from the R domain of diphtheria toxin for the treatment of diseases associated with the activation of the HB-EGF/EGFR pathway

Abstract

A ligand recombinant protein inhibiting HB-EGF (Heparin-Binding Epidermal Growth Factor like), from the R domain of diphtheria toxin, which can be used for the treatment and diagnosis of diseases involving the activation of the HB-EGF/EGFR pathway.

Claims

1. A recombinant protein having at least 90% sequence identity with residues 380 to 535 of the amino acid sequence SEQ ID NO: 1, wherein the recombinant protein comprises at least the substitutions: (i) Y380K or Y380E, and (ii) L390T, relative to SEQ ID NO: 1.

2. The protein of claim 1, comprising at least substitutions Y380K and L390T relative to SEQ ID NO: 1.

3. The protein of claim 2, further comprising at least a substitution Q387E relative to SEQ ID NO: 1.

4. The protein of claim 1, further comprising a substitution A395T relative to SEQ ID NO: 1.

5. The protein of claim 1, further comprising at least one substitution relative to SEQ ID NO: 1 selected from the group consisting of F389Y and G510A.

6. The protein of claim 1, further comprising at least one substitution relative to SEQ ID NO: 1 selected from the group consisting of: N399K, V452T, T517E, V483Q, H492E, S494K, T436H and E497D.

7. The protein of claim 1, comprising substitutions N399K, V452T, T517E, V483Q, H492E and S494K relative to SEQ ID NO: 1.

8. The protein of claim 1, comprising one of amino acid sequences SEQ ID NOs: 2 to 9.

9. The protein of claim 1, which is labeled with a detectable tracer.

10. The protein of claim 9, wherein the detectable tracer is a radioactive isotope or a fluorophore.

11. A pharmaceutical composition, comprising at least one recombinant protein of claim 1, and a pharmaceutically acceptable vehicle.

12. A method for treating a disease associated with activation of Heparin-Binding Epidermal Growth Factor (HB-EGF/EGFR) pathway in a subject in need thereof, the method comprising: administering the recombinant protein of claim 1 to the subject, at a therapeutically effective amount, wherein the disease is selected from the group consisting of: rapidly progressive glomerulonephritis, ovarian cancer, endometrial cancer, breast cancer, uterine cancer, bladder cancer, stomach cancer, skin cancer, brain cancer, lung cancer, vasospasm associated with cerebral contusions, cardiac hypertrophy, smooth muscle cell hyperplasia, pulmonary hypertension, diabetic retinopathies and arterial restenosis.

13. A method for an in vitro diagnosis of a disease associated with activation of HB-EGF/EGFR pathway, the method comprising: obtaining a biological sample from a subject, contacting the protein of claim 9 with the biological sample, thereby forming a complex between the protein and HB-EGF or pro-HB-EGF, detecting the presence of the complex, and diagnosing the subject as having the disease where the presence of the complex is detected, wherein the disease is selected from the group consisting of: rapidly progressive glomerulonephritis, ovarian cancer, endometrial cancer, breast cancer, uterine cancer, bladder cancer, stomach cancer, skin cancer, brain cancer, lung cancer, vasospasm associated with cerebral contusions, cardiac hypertrophy, smooth muscle cell hyperplasia, pulmonary hypertension, diabetic retinopathies and arterial restenosis.

14. The protein of claim 1, comprising at least substitutions Y380E and L390T relative to SEQ ID NO: 1.

Description

(1) In addition to the above arrangements, the invention also comprises other arrangements, which will emerge from the following description, which refers to exemplary embodiments of the subject matter of the present invention, with reference to the appended drawings in which:

(2) FIG. 1 represents the mutations (A) and the analysis (B) of the soluble and insoluble fractions of the 5 clones selected by screening of the R1 library in order to produce a DTR protein which is more soluble than DTR.sub.WT. All the clones are soluble; the best clone is the 1 D6 clone;

(3) FIG. 2 shows the effects of the A395T and N399D mutations on the solubility of the DTR protein produced by the 1D6 clone (denoted here G1). The best clone contains the A395T mutation;

(4) FIG. 3 represents: (A) the inhibition of the toxic effect of increasing doses of diphtheria toxin on Vero cells by increasing doses of DTR1 and (B) the Schild regression for evaluating the Kd of DTR1 for HB-EGF according to the equation log (EC.sub.50−1)=K.sub.d log (B) where EC.sub.50 is the concentration of diphtheria toxin which gives 50% toxicity and B is the concentration of inhibitor DTR1;

(5) FIG. 4 represents the inhibition of the toxic effect of diphtheria toxin at the concentration of 10.sup.−11 M on Vero cells by each DTR1 mutant (denoted here G2) at a concentration of 10.sup.−9 M;

(6) FIG. 5 represents: (A) the inhibition of the toxic effect of increasing doses of diphtheria toxin on Vero cells by increasing doses of DTR8 and (B): the Schild regression for evaluating the Kd of DTR8 for HB-EGF according to the equation log (EC.sub.50−1)=K.sub.d log (B) where EC.sub.50 is the concentration of diphtheria toxin which gives 50% toxicity and B is the concentration of inhibitor DTR8;

(7) FIG. 6 represents the inhibition of the proliferative effect of increasing doses of HB-EGF on Ba/F3 cells expressing the EGFR and the growth of which is HB-EGF-dependent, by increasing doses of DTR8 (A) and of CRM197 (B);

(8) FIG. 7 represents the recognition of the CRM197, diphtheria toxin domain C (C) and domain T (T), DTR1 and DTR8 proteins by the antibodies present in the serum of 20 healthy donors. The titer is defined by the serum dilution which gives a fluorescence signal of 5000 arbitrary units in the ELISA assay, after subtraction of the background noise. A titer ≦20 corresponds to the background noise and therefore to the absence of measurable antibody binding. A titer of 30 was arbitrarily set for distinguishing weak antibody responses from medium or strong responses.

EXAMPLE 1

Materials and Methods

(9) 1) Cloning of Genetic Sequences, Expression and Purification of the DTR.sub.WT Protein and of its Mutants

(10) The genetic sequence encoding the DTR domain of diphtheria toxin, covering the sequence Y380-S535 of the native toxin, was the starting point for this study. All the genetic sequences were synthesized, after optimization for expression in E. coli, by the company Geneart according to the previously determined protein sequences. These sequences were cloned into the pET28a(+) vector (Novagen) at the NcoI and SalI restriction sites. The presence of the NcoI site generates the non-native codons M and G corresponding to the N-terminal end of the recombinant protein. The sequence SEQ ID NO: 16 is the optimized nucleotide sequence which encodes the wild-type DTR protein (DTR.sub.WT) of 158 amino acids, which consists of the residues M and G followed by the residues Y380 to S535 of native diphtheria toxin. The optimized nucleotide sequences encoding the mutant and soluble forms of the DTR protein derive from the sequence SEQ ID NO: 16 by replacement of each codon to be mutated with an optimized codon encoding the mutated amino acid, as indicated in Table I:

(11) TABLE-US-00001 TABLE I List of the optimized codons chosen for the DTR mutations Wild- Optimized type mutated Mutation codon codon Y380K tac aaa Y380E tac gaa P382T ccg acg Q387E cag gag Q387K cag aag P388T ccg acg F389Y ttt tat L390T ctg acc L390N ctg aac H391K cat aaa A395T gcg acc N399D aac gat N399K aac aaa V401Q gtg cag L427Q ctg cag L427N ctg aac L427S ctg agc T436K acc aaa T436H acc cat V452T gtg acg I457D att gat I457E att gaa R460T cgt acc A463T gcg acc A463S gcg agc A463E gcg gaa A463D gcg gat A463G gcg ggc Y478T tat acc V483D gtg gat V483E gtg gaa V483H gtg cat V483Q gtg cag A490G gcg ggc H492E cat gaa S494K agc aaa S496K agc aaa E497D gaa gat G510A ggc gcg G510M ggc atg G510Q ggc cag G510S ggc agc Q515E cag gaa T517D acc gat T517E acc gaa T521R acc cgc K522R aaa cgc

(12) The expression of the DTR.sub.WT protein was carried out in the Escherichia coli BL21(DE3) bacterium in Terrific Broth medium in the presence of 50 μg/ml of kanamycin at 37° C. The induction of the protein is carried out by adding 1 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG). The bacteria were recovered by centrifugation at 5000 g for 45 min and lyzed in a buffer of 20 mM sodium phosphate, 500 mM NaCl, 0.25 mM phenylmethylsulfonyl fluoride (PMSF), lysozyme (0.25 mg/ml), pH 8, by passing them through a Cell Disrupter (Constant Systems). The inclusion bodies containing the protein were solubilized in a solution containing 8 M of urea in 0.1 M Tris-HCl, pH 8. The protein is purified by cation exchange chromatography on a 5 ml HiTrap™ SP column (GE Health Care Life Sciences) according to a buffer gradient between 8 M of urea in 0.1 M Tris-HCl, pH 8, and 8 M of urea, 1 M NaCl in 0.1 M Tris-HCl, pH 8, and then folded by means of dialysis in 2 steps, firstly against a buffer of 20 mM Tris-HCl, 50 mM NaCl, 1 mM cysteine/8 mM cystine, 0.5% sarkosyl, pH 8, and then against a buffer of 20 mM Tris-HCl, 50 mM NaCl, 0.5% sarkosyl, pH 8.

(13) The expression of the mutant and soluble forms of the DTR protein was carried out in the Escherichia coli BL21(DE3) bacterium in 2YT medium in the presence of 50 μg/mi of kanamycin. After 2 h 20 min of culture at 37° C., the induction of the protein is carried out by adding 1 mM of IPTG. Then, after 4 h 30 min at 30° C., the bacteria were recovered by centrifugation at 5000 g for 30 min and lyzed in a buffer of 20 mM sodium phosphate, 0.25 mM PMSF, 2 mM MgCl.sub.2, supplemented with benzonase (6.25 U/ml), pH 7.8, in a Cell Disrupter (Constant Systems). The protein, which is soluble in the lysis mixture, was purified by cation exchange chromatography on a 5 ml HiTrap™ SP column according to a gradient between 20 mM sodium phosphate, pH 7.8, and 20 mM sodium phosphate, 1 M NaCl, pH 7.8 (GE), and stored at −20° C. in a buffer of 20 mM sodium phosphate, 500 mM NaCl, pH 7.8.

(14) 2) Screen for Selecting Soluble Mutants

(15) Three DTR sequence libraries produced by synthesis and cloned into the pET28a(+) plasmid were transfected into the E. coli strain BL21(DE3) (Geneart). For each library, clones were subcultured in 2 or 4 96-well plates in culture medium in the presence of a selection antibiotic and of expression inducer (IPTG). After growth, the bacteria were lyzed using a Bugbuster™ solution (0.5× final concentration) (Novagen) with the addition of Lysonase™ (Novagen). After centrifugation of the plates for 20 min at 2700 g at 4° C., the presence of inclusion bodies was evaluated well-by-well through the size of the pellet, and the soluble protein fraction was identified by analysis of the lysis supernatants on a polyacrylamide gel under denaturing conditions and with Coomassie blue staining.

(16) 3) Test for Activity of the DTR Proteins by Inhibition of Diphtheria Toxin Toxicity

(17) Vero cells (ATCC CCL-81™) were seeded into 96-well Cytostar-T™ scintillating plates (Perkin Elmer) at 50 000 cells per well in DMEM medium (Dulbecco's modified Eagle's minimum essential medium) supplemented with 2 mM of glutamine, 1% of a penicillin/streptomycin solution, 1% of a solution of nonessential amino acids and 10% of fetal calf serium. Variable concentrations of diphtheria toxin (Sigma) were added in duplicates. These conditions were repeated in the presence of the antagonist tested: DTR.sub.WT, mutant DTR or CRM197 protein (Sigma). After incubation for 22 h at 37° C. in a 5% CO.sub.2 atmosphere, the culture medium was replaced with leucine-free medium containing 1 μCi/well of .sup.14[C]-leucine (GE Health Care Life Sciences). After incubation for 5 h at 37° C. in a 5% CO.sub.2 atmosphere, the radioactivity incorporated into the cells was counted by placing the plates, covered with an adhesive film, in a MicroBeta® apparatus (Wallac).

(18) 4) Test for Activity of the DTR Proteins by Inhibition of the Proliferating Activity of HB-EGF

(19) The test uses a murine lymphoid cell line Ba/F3 (Palacios, R & Steinmetz, M., Cell, 1985, 81, 727-734) transfected in the laboratory with the EGFR gene. This line is dependent on HB-EGF or on amphiregulin for its growth. The cells were seeded into 96-well Nunclon™ plates (Nunc) at 10 000 cells per well in RPMI (Roswell Park Memorial Institute) 1640 medium supplemented with 2 mM of glutamine, 1% of a penicillin/streptomycin solution, 1% of a solution of nonessential amino acids and 10% of fetal calf serum. Variable concentrations of HB-EGF or of amphiregulin were added in duplicate. These conditions were repeated in the presence of the antagonist tested: mutant DTR or CRM197 protein. After incubation for 24 h at 37° C. in a 5% CO.sub.2 atmosphere, 1 μCi/well of .sup.3[H]-thymidine (GE Health Care Life Sciences) was added to each well. After incubation for 5 h at 37° C. in a 5% CO.sub.2 atmosphere, the cells were aspirated on glass fiber filter (filtermat A, Wallac) using a Tomtec® apparatus. After drying of the filters, the latter were placed in a sealed bag in the presence of scintillation fluid and the radioactivity incorporated into the cells was counted in a MicroBeta® apparatus (Wallac).

(20) 5) Identification of CD4 T Epitopes of the DTR Protein

(21) The identification of the CD4 T epitopes of the DTR protein is carried out by analysis of the DTR peptides recognized by DTR-specific CD4 T lymphocyte lines, restricted with respect to the HLA-DRB1 molecules which are predominant in caucasian phenotypes. The protocols used are those previously described in application WO 2010/076413 with the exception of the following modifications: (1) the DTR-specific CD4 T lympocyte lines are produced by coculturing of donor CD4 T lymphocytes with autologous mature dendritic cells loaded with a pool of overlapping DTR peptides, and (2) the specificity of the lines produced is analyzed by means of an ELISPOT-IFN-γ assay using autologous peripheral blood mononuclear cells (PBMCs) loaded either with the peptide pool used to produce the CD4 T lymphocyte line, in order to verify the specificity of the line for DTR, or with a peptide of the pool, in order to determine the specificity of each line.

(22) a) Isolation of the CD4 T Lymphocytes from PBMCs

(23) Seven donors of different age and different HLA-DRB1 phenotype were selected such that all of these donors express the 8 HLA-DRB1 molecules which are predominant in caucasian phenotypes (HLA-DR1; HLA-DR3; HLA-DR4; HLA-DR7; HLA-DR11; HLA-DR13; HLA-DR15; HLA-DR8). The CD4 T lymphocytes of each donor were isolated from the peripheral blood mononuclear cells (PBMCs), with a degree of purity greater than 98%, by magnetic sorting on a column using magnetic beads coupled to an anti-CD4 antibody, according to a standard procedure defined by the manufacturer (Myltenyi Biotech).

(24) b) Production and Characterization of the DTR-Specific CD4 T Lymphocyte Lines

(25) 25 overlapping peptides of 15 amino acids covering the entire DTR sequence was synthesized (Intavis Bioanalytical Instruments).

(26) TABLE-US-00002 TABLE II Overlapping peptides covering    the DTR sequence (SEQ ID NOs: 18 to 42) SEQ Locali- ID Peptide zation Amino acid sequence NO:  1 378-392 M G Y S P G H K T Q P F L H D 18  2 385-399 K T Q P F L H D G Y A V S W N 19  3 391-405 H D G Y A V S W N T V E D S I 20  4 397-411 S W N T V E D S I I R T G F Q 21  5 403-417 D S I I R T G F Q G E S G H D 22  6 409-423 G F Q G E S G H D I K I T A E 23  7 415-429 G H D I K I T A E N T P L P I 24  8 421-435 T A E N T P L P I A G V L L P 25  9 427-441 L P I A G V L L P T I P G K L 26 10 433-447 L L P T I P G K L D V N K S K 27 11 439-453 G K L D V N K S K T H I S V N 28 12 445-459 K S K T H I S V N G R K I R M 29 13 451-465 S V N G R K I R M R C R A I D 30 14 457-471 I R M R C R A I D G D V T F C 31 15 463-477 A I D G D V T F C R P K S P V 32 16 469-483 T F C R P K S P V Y V G N G V 33 17 475-489 S P V Y V G N G V H A N L H V 34 18 482-496 G V H A N L H V A F H R S S S 35 19 488-502 H V A F H R S S S E K I H S N 36 20 494-508 S S S E K I H S N E I S S D S 37 21 500-514 H S N E I S S D S I G V L G Y 38 22 506-520 S D S I G V L G Y Q K T V D H 39 23 512-526 L G Y Q K T V D H T K V N S K 40 24 518-532 V D H T K V N S K L S L F F E 41 25 521-535 T K V N S K L S L F F E I K S 42

(27) The peptides were grouped together in 3 pools: Pool 1: peptides 1 to 8 Pool 2: peptides 9 to 16 Pool 3: peptides 17 to 25.

(28) Each of the pools is used in vitro to repeatedly independently sensitize CD4 T lymphocytes from the donors selected for the study. A minimum of 30 coculture wells are produced per pool of peptides. Firstly, the sensitized CD4 T cells capable of recognizing the peptide pool used during the stimulation, called CD4 T lymphocyte lines, are selected by means of an ELISPOT-IFN-γ assay using autologous PBMCs loaded with the peptide pool, as antigen-presenting cells. Secondly, the peptide(s) specifically recognized by the CD4 T lines selected during the first analysis are identified by means of an ELISPOT-IFN-γ assay using autologous PBMCs separately loaded with each of the peptides of the pool, as antigen-presenting cells. The antigen (isolated peptide or peptide pool) recognition specificity is defined by: (1) a ratio between the number of CD4 T lymphocytes producing IFN-γ in response to the antigen (PBMCs+peptides) compared with the background noise (absence of the antigen, i.e. PBMCs alone) which is greater than 2, and (2) a minimum number of 30 spots in the presence of antigen once the background noise has been subtracted.

(29) 6) Evaluation of the Antigenicity of the DTR Proteins

(30) At each step, the incubations were carried out under one or other of the following three conditions: 1 h at 37° C. or 2 h at 20° C. or 16 h at 4° C. The proteins tested (CRM197, catalytic domain, translocation domain, DTR1 and DTR8) were solubilized at the concentration of 1.7×10.sup.−8 M in the PBS buffer at pH 7.4 so as to be adsorbed onto 96-well Maxisorp™ plates (Nunc). The wells were saturated with a solution of bovine serum albumin (BSA) (Sigma) at 3% in PBS. After 4 washes with a solution of PBS buffer containing 0.05% of Tween 20, the healthy donor sera were incubated in duplicate in the wells after dilution to 1/20.sup.th, 1/200.sup.th or 1/2000.sup.th in a PBS buffer containing 0.2% of BSA and 0.05% of Tween 20. After 4 washes with a solution of PBS buffer containing 0.05% de Tween 20, a goat anti-human IgG antibody conjugated to alkaline phosphatase (Sigma) was incubated in the wells after dilution to 1/200.sup.th in a solution of PBS containing 0.2% of BSA and 0.05% of Tween 20. After 4 washes in a solution of PBS containing 0.05% of Tween 20, the test was visualized by incubation in a 0.1 mM solution of 4-methylumbelliferyl phosphate (Sigma) diluted in the buffer of 50 mM carbonate, 1 mM MgCl.sub.2 at pH 9.8 for 30 min at 20° C. The fluorescence of the wells (emission at 450 nm) was measured in a Victor fluorimeter (Wallac) by excitation at 365 nm.

(31) Before being tested, the sera were left to stand at 20° C. for a day, and centrifuged at 10 000 g for 10 min, and then 0.003% of thimerosal was added before storage at 4° C. or at −20° C.

EXAMPLE 2

Improvement of the Solubility of the DTR Protein

(32) The DTR.sub.WT protein expressed in E. coli accumulates in insoluble inclusion bodies. The protein could be obtained, solubilized and purified only in the presence of 0.5% of sarkosyl or of sodium dodecyl sulfate, which are detergents that are incompatible with therapeutic use at these concentrations. The use of other solubilizing molecules was unsuccessful (Tween-80, sucrose, arginine). The use of chaotropic agents such as urea or guanidine chloride for solubilizing the protein, followed by dialysis against various folding buffers, also does not make it possible to obtain a soluble functional protein. The influence of the DTR truncation site relative to the complete diphtheria toxin sequence was also studied. The DTR forms beginning at residue A379, Y380, S381, P382, G383, H384 or K385 were all insoluble in the absence of detergent.

(33) The strategy used to increase the solubility of the DTR protein consisted in mutating the hydrophobic residues present at the surface of the protein with polar or charged hydrophilic residues. This is because the hydrophobic residues at the surface of a protein are potentially responsible for low solubility and for a tendency to aggregate. The mutations to be introduced were identified on the basis of molecular modeling. The potential effect of the mutations on the structure of the protein is indicated in Table III.

(34) TABLE-US-00003 TABLE III Expected effect of the selected mutations Posi- Muta- tions tions Observation regarding structure and interactions Y380 — Flexible at the end of the N-ter loop, no stable hydrogen bond Y380E Ionic bond with K385 Y380K Ionic bond with E532 P382 — In a type II turn P382T Q387 — In N-ter loop, no stable hydrogen bond Q387E Ionic bond with Y380K P388 — In N-ter loop, just before a beta strand P388T L390 — In a beta strand L390N Donor of hydrogen bond for the CO group of the backbone of Y394 L390T Beta-branched residue favored A395 — In a beta strand A395T Beta-branched residue favored N399 — In a loop, no stable hydrogen bond N399D Ionic bond with K419 N424 — In a loop, no stable hydrogen bond N424D Ionic bond with N481K N424E Ionic bond with N481K N424K Ionic bond with E423 or N481E or N481D P426 — At the end of a loop P426T L427 — In a beta strand, Van der Waals contacts with Y394 L427K Van der Waals contacts with Y394, donor of hydrogen bond for T425, ionic bond with E423 L427R Van der Waals contacts with Y394, donor of hydrogen bond for T425, ionic bond with E423 P428 — In a bulge P428T P476 — In a bulge P476T Y478 — In a beta strand, Van der Waals contacts with P426, P428, P476 Y478D Y478N N481 — In a loop, no stable hydrogen bond N481D Ionic bond with N424K N481E Ionic bond with N424K N481K Ionic bond with N424E or N424D V483 — In a loop V483T

(35) Three DTR DNA sequence libraries were prepared by synthesis (Table IV). Each library contained sequences mutated on 4 or 5 codons chosen according to the molecular modeling data. Each library corresponded to codons mutated in the same region of the coding sequence. The sequences contained partial degeneracies of the codons to be mutated so as to limit the possible mutations to potentially acceptable hydrophilic residues according to the modeling data (1, 2 or 3 possible mutations per position). The possibility of retaining the wild-type codon was maintained, in the case where the position tested cannot tolerate a mutation (Table IV), except for Y380, the N-terminal position of which, which is relatively not very constrained, was considered to be tolerant.

(36) TABLE-US-00004 TABLE IV Combinations of possible mutations expected for each of the three libraries of mutant sequences generated to increase the solubility of the R domain (R1, R2, R3) Mutated Mutated Possible Library region positions residues diversity R1 Y380 K/E 72 P382 P/T Q387 Q/E/K P388 P/T L390 L/T/N R2 N424 N/K/D/E 48 P426 P/T L427 L/R/K P428 P/T R3 P476 P/T 48 Y478 Y/N/D N481 N/K/D/E V483 V/T *The diversity indicates the number of combinations of mutations possible for each of the libraries.

(37) After transfection of the DNA libraries, the bacterial colonies obtained, each corresponding to a given mutation combination, were analyzed for their possible capacity to produce a soluble mutant form of the DTR protein. For this, 740 clones derived from the three libraries, representing 85% of the expected diversity (41 different clones for each library), were subcultured in 96-well plates, cultured under protein expression induction conditions, centrifuged, and then lyzed in a lysis solution. After centrifugation, the solubility of the proteins expressed by each clone was evaluated by observation for a possible inclusion body pellet at the bottom of the well and by analysis of the supernatant by polyacrylamide gel electrophoresis under denaturing conditions and with Coomassie blue staining (FIG. 1).

(38) The analysis of the colonies resulted in selecting the clones where there was an absence of pellet (pellet corresponding to the inclusion bodies) or a pellet of reduced size and also a large amount of protein in the lysis supernatant. Only the R1 library made it possible to generate clones producing soluble DTR (FIG. 1), there being five of said clones. These clones, called 1D3, 1D6, 2C8, 5G5 and 1H3, comprise the nucleotide sequences encoding the proteins SEQ ID NOs: 2 to 6, respectively. All the clones were soluble after purification according to the method described in Example 1. The 1D6 clone, also called G1, which makes it possible to obtain the largest amount of DTR in soluble form after expression at 30° C., carried the mutations: Y380K/Q387E/L390T.

(39) The modeling approach led to the proposing of two additional mutations, not represented in the R1 library: A395T and N399D. These two mutations were each introduced into the 1D6 clone (or G1 clone) in order to search for an increased solubilizing effect, which was the case for the A395T mutation (FIG. 2).

(40) In total, the most soluble DTR protein, called DTR1 (SEQ ID NO: 7), carries the mutations: Y380K/Q387E/L390T/A395T.

(41) This does not rule out the fact that other DTR mutations may further improve its solubility and its production in E. coli.

(42) The HB-EGF-binding activity of the DTR1 protein was evaluated through its capacity to inhibit the poisoning of Vero cells by diphtheria toxin, and therefore the binding of the toxin to pro-HB-EGF. The results show that DTR1 inhibits the poisoning of Vero cells by diphtheria toxin in a dose-dependent manner (FIG. 3). The inhibitory effect of DTR1 was compared with that of CRM197, of the 1D6 clone and of DTR.sub.WT solubilized in 0.5% of sarkosyl. The Kd values estimated by Schild regression for the three interactions give the following values:

(43) TABLE-US-00005 TABLE V Affinity for HB-EGF Protein Kd (pM) DTR1 ~49 Clone 1D6 ~25 DTR.sub.WT (+0.5% sarkosyl) ~6500 CRM197 ~3100

(44) Biacore experiments in which HB-EGF was immobilized on the chip of the apparatus and DTR1 was injected into the mobile fraction in order to measure the association and dissociation constants made it possible to confirm the estimated Kd of CRM197 and to show a much higher Kd for DTR1. However, the slowness of the dissociation does not allow an accurate measurement of the dissociation constant of DTR1 in Biacore. In the rest of the study, the affinity of the mutants was estimated by cytotoxicity and S child regression.

EXAMPLE 3

Improvement of the Binding Site of the DTR Protein

(45) The structure of diphtheria toxin in interaction with HB-EGF (Louie et al., Mol. Cell., 1997, 1, 67-78) makes it possible to analyze the interface between the two proteins. This structural analysis, coupled with in silico molecular modeling experiments, made it possible to select 10 mutations in the DTR binding site in order to increase the affinity of the protein for HB-EGF. These mutations were intended to increase the enthalpy of the interaction by promoting the formation of hydrogen bonds, of salt bridges and/or of Van der Waals contacts between the two proteins. The potential effect of the selected mutations on the structure of the protein is indicated in Table VI.

(46) TABLE-US-00006 TABLE VI Observations of structural nature regarding the residues selected for increasing the affinity of the DTR protein and expected effect of the mutations Positions Mutations Observations of structural nature and potential interactions F389 — In a beta strand, at the edge of the region of interaction with HB-EGF, para-position close to a water molecule of the structured interface and HB-EGF H139 F389Y Donor or acceptor of hydrogen bond for a water molecule of the structured interface or HB-EGF H139 H391 — In a beta strand, at the edge of the region of interaction with HB-EGF, donor of hydrogen bond for HB-EGF E141 H391K Ionic bond with HB-EGF E141 G510 — In a beta strand, at the center of the region of interaction with HB-EGF, next to a cavity between DTR and HB-EGF, opposite the CO group of the backbone of HB-EGF C132 G510A Most conservative change for filling the cavity and increasing Van der Waals contacts G510M Flexible hydrophobic side chain for filling the cavity and increasing Van der Waals contacts G510Q Donor of hydrogen bond for the CO group of the backbone of HB-EGF C132 and/or acceptor of hydrogen H bond for the NH group of the backbone of HB-EGF C134. G510S Small polar side chain for filling the cavity and increasing Van der Waals contacts T521 — In a type II turn-like structure, at the edge of the region of interaction with HB-EGF T521R Donor of hydrogen bond for the CO group of the backbone of HB-EGF K111 and/or the CO group of the backbone of HB-EGF K113 Q515 — In a beta strand, at the edge of the region of interaction with HB-EGF, donor of hydrogen bond for the CO group of the backbone of HB-EGF L127 Q515E In an ionic and hydrogen bond network involving K522R, HB-EGF R128, the CO group of the backbone of HB-EGF R128, the CO group of the backbone of HB-EGF L127 K522 — In a beta strand, at the edge of the region of interaction with HB-EGF, T517 hydrogen bond donor K522R In an ionic and hydrogen bond network involving Q515E, HB-EGF R128, the CO group of the backbone of HB-EGF R128, the CO group of the backbone of HB-EGF L127

(47) The mutations were introduced into the DTR1 protein by site-directed mutagenesis (Table VII).

(48) TABLE-US-00007 TABLE VII Mutations for improving the HB-EGF-binding affinity of DTR identified by molecular modeling and tested experimentally F389Y G510Q H391K G510S F389Y/H391K T521R G510A Q515E/K522R G510M F389Y/H391K/Q515E/K522R

(49) The proteins were expressed in E. coli. They were tested for their capacity to inhibit the binding of diphtheria toxin to pro-HB-EGF according to the cytotoxicity test described in Example 1. FIG. 4A shows the capacity of each mutant to possibly inhibit the toxicity of diphtheria toxin on Vero cells. The results show that only the F389Y and G510A mutants significantly inhibit the toxicity of diphtheria toxin. The introduction of these two mutations into DTR1 leads to a cumulative effect (FIG. 4B).

(50) The most active mutant, corresponding to the DTR1 protein carrying the F389Y and G510A mutations, was called DTR3 (SEQ ID NO: 8). Its HB-EGF-binding affinity was evaluated by the capacity of increasing doses of DTR3 to inhibit the toxic effect of increasing doses of diphtheria toxin on Vero cells as described in Example 2. The Kd estimated by Schild regression from the EC50 values of the inhibition curves is given in Table VIII.

(51) TABLE-US-00008 TABLE VIII Affinity for HB-EGF Protein Kd (pM) DTR3 ~9.5

(52) This value suggests that DTR3 has an affinity for HB-EGF which is at least 300 times greater than CRM197 and 5 times greater than DTR1.

EXAMPLE 4

Decrease in the Immunogenicity of the DTR Protein by Elimination of the Main CD4 T Epitopes

(53) The capacity of 25 overlapping peptides covering the entire sequence of DTR.sub.WT to activate specific CD4 T lymphocytes was tested by ELISPOT, in vitro immunization experiments using CD4 T lymphocytes and dendritic cells purified from the blood of 7 healthy donors, of different age and of different HLA-DRB 1 phenotype.

(54) The results show that the immune response against the DTR.sub.WT protein is directed predominantly against five epitope regions covering 60% of the protein and against which at least 71% of the donors responded, i.e. 5 donors out of 7 studied (Table IX): the L.sub.427-L.sub.441 region (peptide 9) against which specific CD4 T lymphocytes were detected for all of the donors studied with a high magnitude (51 lines among 230 lines screened, i.e. 22%), the H.sub.391-Q.sub.411/S.sub.451-D.sub.465/S.sub.475-N.sub.502 regions (peptides 3-4, 13 and 17-18-19 respectively) against which, overall, specific CD4 T cells were detected for 86% of the donors (i.e. 6 donors/7) with a slightly more moderate magnitude than for the region described above (20 to 30 specific CD4 T lymphocyte lines, i.e. 8.5 to 13%), and the S.sub.506-H.sub.520 region (peptide 22) against which specific CD4 T cells were detected for 71% of the donors with a magnitude of 8%.

(55) TABLE-US-00009 TABLE IX Results, by peptide and by donor, for the CD4 T lymphocyte lines specific for the DTR protein Donors P668 P661 P659 P663 P664 P667 P658 HLA-DRB1 DR3 DR1 DR13 DR4 DR1 DR8 DR3 *Total DR13 DR11 DR7 DR11 DR4 DR15 DR7 number of Age CD4 T Responder Peptides 52 44 29 62 57 44 30 lines frequency 1 378-392 1 1 14 2 385-399 1 4 2 7 43 3 391-405 1 3 2 1 4 11 71 4 397-411 1 3 1 1 13 8 27 86 5 403-417 1 1 14 6 409-423 1 6 7 29 7 415-429 1 1 14 8 421-435 1 1 2 29 9 427-441 3 3 5 3 13 4 20 51 100 10 433-447 1 2 1 18 22 57 11 439-453 2 2 1 5 43 12 445-459 9 5 3 17 43 13 451-465 9 14 1 2 3 1 30 86 14 457-471 2 1 1 4 43 15 463-477 10 10 14 16 469-483 2 1 3 29 17 475-489 7 2 8 5 1 21 71 18-19 482-502 4 1 5 4 1 3 18 86 20 494-508 2 2 4 29 21 500-514 1 1 4 6 43 22 506-520 1 2 6 4 5 18 71 23 512-526 3 2 5 29 24 518-532 2 1 3 29 25 521-535 1 1 14 Total 27 16 44 23 28 30 40 % 30 13 49 19 31 33 44 *Total number of CD4 T lymphocyte lines specific for a peptide among the 230 lines screened

(56) Since 5 epitope regions were identified, it is possible to envision the mutation of these epitopes in order to reduce their binding to HLA-II molecules. The ProPred server was used to predict, within these five regions of DTR.sub.WT, the sequences capable of binding the 8 HLA class II alleles which are the most common in the population (Table X). The same analysis applied to the mutated sequences of DTR1 shows that the mutations introduced in order to improve the solubility of DTR reduced the immunogenicity of the protein (epitopes predicted for the region 378-403). Indeed, the epitope predicted to bind DRB1_0301 disappears and one of the epitopes predicted to bind DRB1_0401 experiences an increase in its threshold, that is to say a decrease in its predicted affinity for the HLA molecule.

(57) TABLE-US-00010 TABLE X Prediction of the sequences capable of binding the 8 HLA class II alleles which are the most common (DRB1_0101, DRB1_0301, DRB1_0401, DRB1_0701, DRB1_1101, DRB1_1301, DRB1_1501, DRB5_0101) within the 5 regions identified as immunogenic by means of a CD4 T lymphocyte activation assay. The peptide sequences correspond to the sequences SEQ ID NOs: 43 to 110. The second column of the table gives the region studied. The first column indicates the mutations proposed for inhibiting the binding of the predicted sequences to the HLA class II molecules. The subsequent columns indicate, for each HLA allele considered, the presence or absence of sequence capable of binding this HLA molecule. Each sequence is preceded by a threshold value reflected in the binding strength: 1/2 (strong binding), 3/4 (medium binding), 5/6 (weak binding). The residues in bold correspond to the mutations proposed for abolishing the binding of the sequences to the HLA allele considered. Variant DRB1_0101 DRB1_0301 DRB1_0401 DRB1_0701 Sequence (378-403) WT MGYSPGHKTQPFLHDGYAVSWNTVED — 6: FLHDGYAVS 3: FLHDGYAVS — (SEQ ID NO: 43) (SEQ ID NO: 45) (SEQ ID NO: 45) 5: YAVSWNTVE (SEQ ID NO: 46) DTR1 MGKSPGHKTEPFTHDGYTVSWNTVED — — 5: FTHDGYTVS — (SEQ ID NO: 44) (SEQ ID NO: 47) 5: YTVSWNTVE (SEQ ID NO: 48) Sequence (391-411) DTR1 HDGYTVSWNTVEDSIIRTGFQ — — 4: VSWNTVEDS 6: WNTVEDSII (SEQ ID NO: 49) (SEQ ID NO: 53) (SEQ ID NO: 55) 5: YTVSWNTVE (SEQ ID NO: 54) N399K HDGYTVSWKTVEDSIITGFQ — — — 5: WKTVEDSII (SEQ ID NO: 50) (SEQ ID NO: 56) V401Q HDGYTVSWNTQEDSIIRTGFQ — — 5: YTVSWNTQE — (SEQ ID NO: 51) (SEQ ID NO: 58) N399K HDGYTVSWKTQEDSIIRTGFQ — — — — V401Q (SEQ ID NO: 52) Sequence (427-441) DTR1 LPIAGVLLPTIPGKL — 6: VLLPTIPGK — 3: LLPTIPGKL (SEQ ID NO: 59) (SEQ ID NO: 63) (SEQ ID NO: 65) L427Q QPIAGVLLPTIPGKL — 6: VLLPTIPGK — 3: LLPTIPGKL (SEQ ID NO: 60) (SEQ ID NO: 63) (SEQ ID NO: 65) T436K LPIAGVLLPKIPGKL — 6: VLLPKIPGK — — (SEQ ID NO: 61) (SEQ ID NO: 64) L427Q QPIAGVLLPKIPGKL — 6: VLLPKIPGK — — T436K (SEQ ID NO: 62) (SEQ ID NO: 64) Sequence (451-465) DTR1 SVNGRKIRMRCRAID — 3: IRMRCRAID — — (SEQ ID NO: 67) (SEQ ID NO: 74) I457D SVNGRKDRMRCRAID — — — — (SEQ ID NO: 68) I457E SVNGRKERMRCRAID — — — — (SEQ ID NO: 69) V452T STNGRKIRMRCRAID — 3: IRMRCRAID — — (SEQ ID NO: 70) (SEQ ID NO: 74) R460T SVNGRKIRMTCRAID — 3: IRMTCRAID — — (SEQ ID NO: 71) (SEQ ID NO: 75) A463D SVNGRKIRMRCRDID — 5: IRMRCRDID — — (SEQ ID NO: 72) (SEQ ID NO: 76) V452T STNGRKIRMTCRDID — 5: IRMTCRDID — — R460T (SEQ ID NO: 73) (SEQ ID NO: 77) A463D Sequence (475-502) DTR1 SPVYVGNGVHANLHVAFHRSSSEKIHSN 3: YVGNGVHAN — 1: YVGNGVHAN 1: FHRSSSEKI (SEQ ID NO: 80) (SEQ ID NO: 91) (SEQ ID NO: 91) (SEQ ID NO: 91) 3: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 92) (SEQ ID NO: 92) 4: VYVGNGVHA (SEQ ID NO: 93) Y478T SPVTVGNGVHANLHVAFHRSSSEKIHSN 3: FHRSSSEKI — 1: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 81) (SEQ ID NO: 92) (SEQ ID NO: 92) (SEQ ID NO: 92) V483D SPVYVGNGDHANLHVAFHRSSSEKIHSN 3: FHRSSSEKI — 1: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 82) (SEQ ID NO: 92) (SEQ ID NO: 92) (SEQ ID NO: 92) 5: YVGNGDHAN (SEQ ID NO: 95) V483E SPVYVGNGEHANLHVAFHRSSSEKIHSN 3: FHRSSSEKI — 1: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 83) (SEQ ID NO: 92) (SEQ ID NO: 92) (SEQ ID NO: 92) V483H SPVYVGNGHHANLHVAFHRSSSEKIHSN 3: FHRSSSEKI — 1: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 84) (SEQ ID NO: 92) (SEQ ID NO: 92) (SEQ ID NO: 92) 5: VYVGNGHHA (SEQ ID NO: 96) V483Q SPVYVGNGQHANLHVAFHRSSSEKIHSN 3: FHRSSSEKI — 1: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 85) (SEQ ID NO: 92) (SEQ ID NO: 92) (SEQ ID NO: 92) 5: VYVGNGQHA (SEQ ID NO: 97) 6: YVGNGQHAN (SEQ ID NO: 98) A490G APVYVGNGVHANLHVGFHRSSSEKIHSN 3: YVGNGVHAN — 1: YVGNGVHAN 1: FHRSSSEKI (SEQ ID NO: 86) (SEQ ID NO: 91) (SEQ ID NO: 91) (SEQ ID NO: 92) 3: FHRSSSEKI 1: FHRSSSEKI (SEQ ID NO: 92) (SEQ ID NO: 92) 4: VYVGNGVHA 6: VGFHRSSSE (SEQ ID NO: 93) (SEQ ID NO: 110) H492E SPVYVGNGVHANLHVAFERSSSEKIHSN 3: YVGNGVHAN — 1: YVGNGVHAN 1: FERSSSEKI (SEQ ID NO: 87) (SEQ ID NO: 91) (SEQ ID NO: 91) (SEQ ID NO: 100) 4: VYVGNGVHA 2: FERSSSEKI (SEQ ID NO: 93) (SEQ ID NO: 100) 5: FERSSSEKI (SEQ ID NO: 100) S494K SPVYVGNGVHANLHVAFHRKSSEKIHSN 3: YVGNGVHAN — 1: YVGNGVHAN 2: FHRKSSEKI (SEQ ID NO: 88) (SEQ ID NO: 91) (SEQ ID NO: 91) (SEQ ID NO: 102) 4: YVYGNGVHA 5: FHRKSSEKI (SEQ ID NO: 93) (SEQ ID NO: 102) S496K SPVYVGNGVHANLHVAFHRSSKEKIHSN 3: YVGNGVHAN 4: FHRSSKEKI 1: YVGNGVHAN 1: FHRSSKEKI (SEQ ID NO: 89) (SEQ ID NO: 91) (SEQ ID NO: 105) (SEQ ID NO: 91) (SEQ ID NO: 105) 4: VYVGNGVHA (SEQ ID NO: 93) Y478T SPVTVGNGVHANLHVGFERKSKEKIHSN — — — — A490G (SEQ ID NO: 90) H492E S494K S496K Sequence (506-520) DTR1 SDSIGVLGYQKTVDH — — 1: LGYQKTVDH — (SEQ ID NO: 106) (SEQ ID NO: 109) T517D SDSIGVLGYQKDVDH — — — — (SEQ ID NO: 107) T517E SDSIGVLGYQKEVDH — — — — (SEQ ID NO: 108) Variant DRB1_1101 DRB1_1301 DRB1_1501 DRB5_0101 Sequence (378-403) WT MGYSPGHKTQPFLHDGYAVSWNTVED — — — — (SEQ ID NO: 43) DTR1 MGKSPGHKTEPFTHDGYTVSWNTVED — — — — (SEQ ID NO: 44) Sequence (391-411) DTR1 HDGYTVSWNTVEDSIIRTGFQ — — — — (SEQ ID NO: 49) N399K HDGYTVSWKTVEDSIITGFQ — 6: VSWKTVEDS — — (SEQ ID NO: 50) (SEQ ID NO: 57) V401Q HDGYTVSWNTQEDSIIRTGFQ — — — — (SEQ ID NO: 51) N399K HDGYTVSWKTQEDSIIRTGFQ — — — — V401Q (SEQ ID NO: 52) Sequence (427-441) DTR1 LPIAGVLLPTIPGKL 3: LPIAGVLLP 4: LPIAGVLLP — 2: LLPTIPGKL (SEQ ID NO: 59) (SEQ ID NO: 66) (SEQ ID NO: 66) (SEQ ID NO: 65) L427Q QPIAGVLLPTIPGKL — — — 2: LLPTIPGKL (SEQ ID NO: 60) (SEQ ID NO: 65) T436K LPIAGVLLPKIPGKL 3: LPIAGVLLP 4: LPIAGVLLP — — (SEQ ID NO: 61) (SEQ ID NO: 66) (SEQ ID NO: 66) L427Q QPIAGVLLPKIPGKL — — — — T436K (SEQ ID NO: 62) Sequence (451-465) DTR1 SVNGRKIRMRCRAID 3: IRMRCRAID 1: VNGRKIRMR 4: IRMRCRAID — (SEQ ID NO: 67) (SEQ ID NO: 74) (SEQ ID NO: 78) (SEQ ID NO: 74) 1: IRMRCRAID (SEQ ID NO: 74) I457D SVNGRKDRMRCRAID — — — — (SEQ ID NO: 68) I457E SVNGRKERMRCRAID — — — — (SEQ ID NO: 69) V452T STNGRKIRMRCRAID 3: IRMRCRAID 1: IRMRCRAID 4: IRMRCRAID — (SEQ ID NO: 70) (SEQ ID NO: 74) (SEQ ID NO: 74) (SEQ ID NO: 74) R460T SVNGRKIRMTCRAID 3: IRMTCRAID 3: VNGRKIRMT 6: IRMTCRAID (SEQ ID NO: 71) (SEQ ID NO: 75) (SEQ ID NO: 79) (SEQ ID NO: 75) 3: IRMTCRAID (SEQ ID NO: 75) A463D SVNGRKIRMRCRDID — 1: VNGRKIRMR — — (SEQ ID NO: 72) (SEQ ID NO: 78) 3: IRMRCRDID (SEQ ID NO: 76) V452T STNGRKIRMTCRDID — — — — R460T (SEQ ID NO: 74) A463D Sequence (475-502) DTR1 SPVYVGNGVHANLHVAFHRSSSEKIHSN 2: YVGNGVHAN 1: LHVAFHRSS — — (SEQ ID NO: 80) (SEQ ID NO: 91) (SEQ ID NO: 94) 5: LHVAFHRSS 4: YVGNGVHAN (SEQ ID NO: 94) (SEQ ID NO: 91) Y478T SPVTVGNGVHANLHVAFHRSSSEKIHSN 5: LHVAFHRSS 1: LHVAFHRSS — — (SEQ ID NO: 81) (SEQ ID NO: 94) (SEQ ID NO: 94) V483D SPVYVGNGDHANLHVAFHRSSSEKIHSN 5: LHVAFHRSS 1: LHVAFHRSS — — (SEQ ID NO: 82) (SEQ ID NO: 94) (SEQ ID NO: 94) V483E SPVYVGNGEHANLHVAFHRSSSEKIHSN 5: LHVAFHRSS 1: LHVAFHRSS — — (SEQ ID NO: 83) (SEQ ID NO: 94) (SEQ ID NO: 94) V483H SPVYVGNGHHANLHVAFHRSSSEKIHSN 5: VYVGNGHHA 1: LHVAFHRSS 6: VYVGNGHHA — (SEQ ID NO: 84) (SEQ ID NO: 96) (SEQ ID NO: 94) (SEQ ID NO: 96) 5: LHVAFHRSS (SEQ ID NO: 94) V483Q SPVYVGNGQHANLHVAFHRSSSEKIHSN 5: YVGNGQHAN 1: LHVAFHRSS — — (SEQ ID NO: 85) (SEQ ID NO: 98) (SEQ ID NO: 94) 5: LHVAFHRSS (SEQ ID NO: 94) A490G APVYVGNGVHANLHVGFHRSSSEKIHSN 2: YVGNGVHAN 4: YVGNGVHAN — — (SEQ ID NO: 86) (SEQ ID NO: 91) (SEQ ID NO: 91) 5: LHVGFHRSS (SEQ ID NO: 99) H492E SPVYVGNGVHANLHVAFERSSSEKIHSN 2: YVGNGVHAN 4: YVGNGVHAN — — (SEQ ID NO: 87) (SEQ ID NO: 91) (SEQ ID NO: 91) 5: LHVAFERSS (SEQ ID NO: 101) S494K SPVYVGNGVHANLHVAFHRKSSEKIHSN 2: YVGNGVHAN 1: LHVAFHRKS — — (SEQ ID NO: 88) (SEQ ID NO: 91) (SEQ ID NO: 103) 5: LHVAFHRKS 4: YVGNGVHAN (SEQ ID NO: 103) (SEQ ID NO: 91) 4: VAFHRKSSE (SEQ ID NO: 104) S496K SPVYVGNGVHANLHVAFHRSSKEKIHSN 2: YVGNGVHAN 1: LHVAFHRSS — — (SEQ ID NO: 89) (SEQ ID NO: 91) (SEQ ID NO: 94) 5: LHVAFHRSS 4: YVGNGVHAN (SEQ ID NO: 94) (SEQ ID NO: 91) 5: FHRSSKEKI (SEQ ID NO: 105) Y478T SPVTVGNGVHANLHVGFERKSKEKIHSN — — — — A490G (SEQ ID NO: 90) H492E S494K S496K Sequence (506-520) DTR1 SDSIGVLGYQKTVDH 6: LGYQKTVDH 5: LGYQKTVDH — — (SEQ ID NO: 106) (SEQ ID NO: 109) (SEQ ID NO: 109) T517D SDSIGVLGYQKDVDH — — — — (SEQ ID NO: 107) T517E SDSIGVLGYQKEVDH — — — — (SEQ ID NO: 108)

(58) The prediction of the anchoring residues of the sequences predicted to bind the HLA class II molecules made it possible to propose a series of mutations indicated in bold in Table X, intended to eliminate potential T epitopes. The mutations were chosen so as to avoid potential destabilization of the protein, which was tested in silico by molecular modeling. Any mutation affecting the HB-EGF-binding site was excluded. The potential effect of the proposed mutations on the structure of the protein is indicated in Table XI.

(59) TABLE-US-00011 TABLE XI Observations of structural nature regarding the residues selected for eliminating CD4 T epitopes of the DTR protein and expected effect of the mutations Positions Mutations Observations of structural nature and potential interactions N399 — Exposed to the solvent, in a loop, no stable hydrogen bond N399K Ionic bond with D417 and donor of hydrogen bond for the CO group of the backbone of N486 V401 — Exposed to the solvent, in a type I turn V401Q Acceptor of hydrogen bond for K385 L427 — Partially buried, in a beta strand, Van der Waals contacts with Y394 L427Q Donor of hydrogen bond for the CO group of the backbone of D392, hydrocarbon-based side chain establishing Van der Waals contacts with Y394 T436 — Partially buried, in a loop, donor of hydrogen bond for the CO group of the backbone of G466, Van der Waals contacts with V443 T436H Acceptor of hydrogen bond for T469, Van der Waals contacts with V443 T436K Ionic bond with A463D I457 — Largely buried; in a beta strand; Van der Waals contacts with I450, V452, P473, V477 I457D Acceptor of hydrogen bond for S475, the NH group of the backbone of S475, the NH group of the backbone of K474 I457E Acceptor of hydrogen bond for S475, the NH group of the backbone of S475, the NH group of the backbone of K474, or the NH group of the backbone of V452 V452 — Partially buried, in a beta strand, Van der Waals contacts by a methyl group of the side chain with I450, I457, V477 V452T Van der Waals contacts by the methyl group of the side chain with I450, I457, V477 A463 — Exposed to the solvent, in a beta strand A463D Ionic bond with T436K R460 — Exposed to the solvent, in a beta strand, is part of a basic patch containing other arginines potentially interacting with heparan sulfate groups bonded to the plasma membrane, close to HB-EGF, donor of hydrogen bond for the CO group of the backbone of P473 R460T Beta-branched residue Y478 — Exposed to the solvent, in a beta strand, Van der Waals contacts with P426, P428, P476 Y478T Beta-branched residue V483 — Exposed to the solvent, in a loop V483D V483E Acceptor of hydrogen bond for V452T, N453 V483H Donor of hydrogen bond for the CO group of the backbone of Y478 V483Q Acceptor or donor of hydrogen bond for V452T, donor of hydrogen bond for the CO group of the backbone of Y478 A490 — Partially buried, in a beta strand A490G H492 — Exposed to the solvent, in a beta strand, acceptor or donor of hydrogen bond for S494 H492E Ionic bond with S494K S494 — Exposed to the solvent, in a loop, acceptor of hydrogen bond for H492 S494K Ionic bond with H492E S496 — Exposed to the solvent, in a loop S496K Donor of hydrogend bond for Q411 T517 — Exposed to the solvent, in a beta strand, acceptor of hydrogen bond for K522 T517D Ionic bond with K522 T517E Ionic bond with K522

(60) The mutations indicated in Table XII were introduced individually, or sometimes in combination, into the sequence of the DTR3 protein, according to the modeling data, then gradually accumulated when they altered neither the production of the recombinant protein nor its biological activity. The biological activity of the mutants was tested by inhibition of the toxicity of diphtheria toxin on Vero cells.

(61) TABLE-US-00012 TABLE XII Mutations introduced into DTR3 in order to eliminate CD4 T epitopes N399K A463S H492E V401Q A463E S494K L427Q A463D S496K L427N A463G H492E/S494K L427S T436K/A463D H492E/S496K T436K V452T/R460T/A463D S494K/S496K T436H Y478T H492E/S494K/S496K V452T V483D E497D I457D V483E T436H/E497D I457E V483H T517D R460T V483Q T517E A463T A490G * In bold, the mutations which do not significantly affect the expression and the activity of the protein, which were therefore retained.

(62) These experiments made it possible to successively retain the following mutations: N399K, V452T, T517E, V483Q, H492E, S494K, T436H and E497D.

(63) The protein derived from DTR3 and accumulating the N399K, V452T, T517E, V483Q, H492E and S494K mutations is called DTR8 (SEQ ID NO: 9); it has an MW of 17458 Da. These mutations made it possible to eliminate 10 of the 26 CD4 T epitopes identified in DTR.sub.WT (Table XIII).

(64) TABLE-US-00013 TABLE XIII Comparison of the number and of the strength of the binding to HLA II of the CD4 T epitopes between DTR.sub.WT and DTR8 Strength of Threshold value CD4 T epitopes binding to reflecting the binding Total DTR HLA II strength 391-411 427-441 451-465 475-502 506-520 number DTR.sub.WT Strong 1 0 0 2 4 1 7 2 0 1 0 1 0 2 Medium 3 0 2 2 2 0 6 4 1 1 1 2 0 5 Weak 5 1 0 0 1 1 3 6 1 1 0 0 1 3 Total 3 5 5 10 3 26 DTR8 Strong 1 0 0 1 0 0 1 2 0 1 0 0 0 1 Medium 3 0 2 2 1 0 5 4 0 1 1 0 0 2 Weak 5 1 0 0 3 0 4 6 1 1 0 1 0 3 Total 2 5 4 5 0 16

(65) Among these 10 epitopes are 7 of the 9 epitopes predicted as being immunodominant epitopes of DTR.sub.WT. The capacity of the resulting protein, DTR8, to induce a CD4-type immune response, i.e. producing antibodies, is thus considerably reduced compared with that of DTR.sub.WT in its wild-type form.

(66) DTR8 contains six more mutations than DTR3. Since any mutation or mutation combination is capable of impairing the function of a protein, it is necessary to evaluate the effect of these additional mutations on the DTR binding affinity for HB-EGF. The HB-EGF-binding activity of the DTR8 protein was evaluated through its capacity to inhibit the poisoning of Vero cells by diphtheria toxin, and therefore the binding of the toxin to pro-HB-EGF. The results (FIG. 5) show that DTR8 inhibits the poisoning of Vero cells by diphtheria toxin in a dose-dependent manner. The inhibitory effect of DTR8 was compared with that of CRM197, DTR.sub.WT solubilized in 0.5% of sarkosyl, DTR3 and DTR1 (FIG. 3). The Kd values estimated by Schild regression for the four interactions give the following values (Table XIV).

(67) TABLE-US-00014 TABLE XIV Affinity of the proteins for HB-EGF Protein Kd (pM) DTR8 2.2 DTR3 9.5 DTR1 49 DTR.sub.WT (+0.5% sarkosyl) 6500 CRM197 3100

(68) Notably and unexpectedly, the addition of the six mutations intended to reduce the immunogenicity of DTR contributed to increasing its affinity for HB-EGF. Overall, DTR8 has approximately 1400 times more affinity for HB-EGF than CRM197.

(69) The HB-EGF-binding activity of the DTR8 protein was also evaluated through its capacity to inhibit the binding of HB-EGF to the EGFR in a Ba/F3 cell line transfected with the EGFR gene and dependent on HB-EGF for its growth. The results (FIG. 6) show that DTR8 inhibits the proliferation of the HB-EGF-dependent cells in a dose-dependent manner. The inhibitory effect of DTR8 was compared with that of CRM197 (FIG. 6). The results show that at least 300 times more CRM197 is necessary in order to obtain an equivalent inhibitory effect. Consequently, DTR8 is at least 300 times more effective than CRM197 in terms of binding to HB-EGF in solution.

(70) In conclusion, these results show that the DTR8 protein is capable of binding to pro-HB-EGF molecules at the surface of cells (FIG. 5) and to HB-EGF in solution (FIG. 6) and of blocking their biological activity. The estimation of the Kd values for the interactions suggests that DTR8 is 1400 times more powerful than CRM197 in terms of binding to HB-EGF.

EXAMPLE 5

Evaluation of the Antigenicity of the DTR1 and DTR8 Proteins

(71) The western population is vaccinated against diphtheria toxin. Individuals who may benefit from treatment with a therapeutic protein of DTR8 type could therefore have antibodies capable of reacting against said protein. The capacity of the sera of 20 healthy donors to recognize diphtheria toxin (in its mutated form CRM197), and its various domains (catalytic (C), translocation (T) and DTR (in the soluble mutated form DTR1)) was compared with the DTR8 protein, by ELISA (FIG. 7). An antibody titer 20 corresponds to the background noise of the assay and therefore to an absence of recognition. A titer threshold at 30 was arbitrarily set in order to distinguish the sera exhibiting a weak or strong response.

(72) The results (FIG. 7) show that 16/20 donors have antibodies against diphtheria toxin. 14/20 donors exhibit a medium to strong antibody response. However, the majority of the antibodies present in each serum considered individually are directed against the C domain of the toxin (FIG. 7). Indeed, 12 sera exhibit a response above the threshold (medium to strong response) and 4 sera exhibit a response below the threshold (weak response). If the reactivity of the sera against the soluble form of the R domain (DTR1) is considered, 15/20 sera recognize the R domain. However, only 7 sera exhibit a response above the threshold (medium to strong response). Notably, the DTR8 protein is considerably less well recognized by the sera of the donors than DTR1. Indeed, only 3/20 sera exhibit medium to strong reactivity against DTR8 and 2 sera exhibit weak reactivity (FIG. 7).

(73) In conclusion, these results show that the antigenicity of the DTR8 protein is weak and considerably reduced compared with that of CRM197. This antigenicity is also reduced in comparison with that of the DTR1 protein, corresponding to the soluble form of DTR carrying the lowest number of mutations. In other words, the mutations introduced into DTR in order to increase its affinity for HB-EGF and to reduce its immunogenicity contribute to considerably reducing its antigenicity.