EP0460021A4

EP0460021A4 - Chimeric toxin

Info

Publication number: EP0460021A4
Application number: EP19900903634
Authority: EP
Inventors: John R. Murphy
Original assignee: Seragen Inc
Current assignee: Seragen Inc
Priority date: 1989-02-23
Filing date: 1990-02-01
Publication date: 1992-08-26
Also published as: WO1990010015A1; EP0460021A1

Abstract

A chimeric toxin composed of a first CD4-derived portion, covalently bonded to a second portion consisting of a portion of diphtheria toxin capable of killing a human T-lymphocyte and incapable of causing generalized binding of said chimeric toxin to human cells.

Description

CHIMERIC TOXIN

Background of the Invention

This invention relates to chimeric toxins useful in the treatment of Acquired Immune Deficiency

Syndrome (AIDS). Chimeric proteins have been constructed by chemically coupling a variety of cell surface directed agents, e.g., monoclonal antibodies, lectins, or hormones, to toxin fragments, e.g., the A chain of the plant toxin ricin, or the A fragment of diphtheria toxin, in order to direct the action of the toxin component toward specific eucaryotic cells. In addition, Murphy U.S. Patent No. 4,675,382, hereby incorporated by reference, used recombinant DNA techniques to produce a hybrid protein, consisting of part of diphtheria toxin linked via a peptide linkage to a cell-specific ligand such as α-melanocyte stimulating hormone, which was selectively toxic for particular target cells, e.g., α-MSH receptor positive human malignant melanoma cells. The CD4 (previously named T4) molecule, which is a surface glycoprotein on a subset of T-lymphocytes, referred to as T4 lymphocytes, is involved in class II (la) MHC recognition and appears to be the physiological receptor for one or more monomorphic regions of class II MHC (Meuer et al., 1982, Proc. Nat. Aca. Sci.

79:4395-4399; Biddison et al. , 1982, J. Exp. Med. , 156:1065-1076; Gay et al. , 1987, Nature 328:626-629). CD4 also functions as a receptor for the exterior envelope glycoprotein (gpi20) of the causative agent of AIDS, human immunodeficiency virus (HIV). It has been shown to be essential for entry of the virus into host cells and for membrane fusion, which contribute to cell-to-cell transmission of the virus and to its cytopathic effects (Sodroski et al. , 1986, Nature 322:470-474; Lifson et al., Nature 323:725-728).

T4 lymphocytes constitute 60-80% of total circulating T-lymphocytes and HIV infection of T4 cells can cause total collapse of the immune system (Curran et al., 1985, Science 229:1352-1357; Weiss et al. , 1986, Nature 324:572-575). Considerable effort has been spent in characterizing the CD4-gpl20 interaction and in trying to interfere with or inhibit this interaction in order to prevent HIV infection of T-cells. Soluble CD4 has been used to attempt to interfere with infection of T-cells by HIV (Hussey et al., 1988, Nature 331:78-81; Fisher et al. , 1988, Nature 331:76-78; Deen et al. , 1988, Nature 331:82-84; Traunecker et al. , 1988, Nature 331:84-86).

Summary of the Invention The invention features a chimeric toxin useful in the treatment of patients infected with HIV. The chimeric toxin of the invention is composed of a first portion consisting of CD4, or a gpl20-binding analog or portion thereof, covalently bonded to a second portion consisting of a portion of diphtheria toxin capable of killing a human T-lymphocyte and incapable of causing generalized binding of the chimeric toxin to human cells.

In preferred embodiments, the toxin of the invention is water-soluble (achieved by using a soluble, gpl20-binding portion of CD4, which does not include the transmembrane region); the first and second portions of the toxin are bonded together by means of a peptide bond, i.e., the toxin is encoded by a fused gene which includes regions encoding both the first and second portions; and the diphtheria toxin derived DNA of the fused gene is located upstream of the CD4-derived DNA regio . In other preferred embodiments, rather than the first and second portions of the chimeric toxin being linked by peptide bond, they are linked by a non-peptide covalent linkage such as a disulfide linkage between a cysteine residue on the first portion of the toxin and a cysteine residue on the second portion. Preferably, these cysteine residues are encoded by cysteine codons on both portions which are not naturally-occurring in the genes encoding diphtheria toxin and CD4. The toxin molecules of the invention can be administered to human patients infected with HIV, operating to treat that infection by one of the following mechanism: the toxin, by virtue of the gpl20-binding portion (the analog of CD4), complexes with cell-free HIV virions in the patient's bloodstream, which in turn direct the toxin to a CD4-bearing T-lymphocyte or macrophage which is a potential target and reservoir of the virus; upon delivery, the potentially infected cell is killed by the diphtheria toxin portion of the chimeric toxin. Alternatively, toxin molecules of the invention could bind gpl20 molecules located on the surface of HIV-infected cells, and enter the cell by membrane turnover, e.g., pinocytosis, or any other naturally occurring cell-entry mechanism. Once inside the HIV-infected cell, the enzymatically active fragment A of diphtheria toxin will interfere with cellular metabolism and result in cell death.

The invention provides a highly specific treatment of HIV infections, without causing harm to cells which are not HIV target cells.

Other features and advantages of the invention will be apparent from the following description of the preferred embodiments thereof, and from the claims. Description of the Preferred Embodiments

The drawings will first briefly be described. Drawings

Fig. 1 is a diagrammatic representation of the CD4 receptor, showing (A) the four domains and the transmembrane binding region ("Tm"), (B) soluble extracellular CD4, and (C) the amino acid sequence of the soluble protein (taken from Clark et al. , 1987 PNAS 84:1649). Fig. 2 is a diagrammatic representation of the diphtheria toxin molecule.

Fig. 3 is a restriction map showing the location and orientation of the diphtheria tox gene on the 3.9 kb BamHI restriction fragment of corynephage beta^tox.

Fig. 4 is the nucleotide sequence of the tox 228 allele and flanking regions, with amino acid residues shown above nucleotides; the tox 228 allele is the same as the wild-type tox allele except for several mutations, notably the presence on the tox 228 allele of an Nrul site (Fig. 4 was adapted from Fig. l of Kaczorek et al. (1983) Science 221, 855).

Figs. 5-6 are diagrammatic representations of the steps involved in the construction of pABC1508, a plasmid which can encode the toxin portion of a chimeric toxin of the invention for disulfide linkage to the CD4 portion, or which can be modified for fusion with CD4-encoding DNA.

Fig. 7 is a diagrammatic representation of the relevant region of pABC1508.

Fig. 8 is a diagrammatic representation of the construction of a plasmid which contains DNA encoding a diphtheria toxin-soluble CD4 chimeric toxin of the invention. Tox Gene

The tox gene, and the diphtheria toxin molecule it encodes, will now briefly be described.

Fig. 1(A) illustrates the four extracellular domains and the transmembrane region of the CD4 receptor.

Fig. 1(B) and (C) illustrate soluble CD4 and its amino acid sequence, respectively. Figs. 2 and 3 illustrate, respectively, the diphtheria toxin molecule and the diphtheria tox gene, located on the 3.9 kb BamHI restriction fragment of corynephage betatox. Fig. 4 gives the sequence of the tox 228 allele.

Referring to Fig. 2, the diphtheria toxin molecule consists of several functional "domains" which can be characterized, starting at the amino terminal end of the molecule, as a hydrophobic signal sequence; enzymatically active Fragment A, the fourteen amino acid exposed protease sensitive disulfide loop (DSL) 1., containing a cleavage domain; Fragment B, which includes the lipid associating regions, e.g., a hydrophilic amphipathic domain and a hydrophobic domain; DSL 1₂; and carboxy terminal end ("a"). DSL 1. contains three ' arginine residues; the Sau3Al site between Fragment A and Fragment B (see Fig. 3) is at a position on the diphtheria toxin gene corresponding to the arginine residue farthest downstream of the three.

The process by which diphtheria toxin intoxicates sensitive eukaryotic cells involves at least the following steps: (i) diphtheria toxin binds to specific receptors on the surface of a sensitive cell; (ii) while bound to its receptor, the toxin molecule is internalized in an endocytic vesicle; (iii) either prior to internalization, or within the endocytic vesicle, the toxin molecule may be cleaved (or processed) at a site in the region of 47,000 daltons from the N-terminal end; (iv) as the pH of the endocytic vesicle decreases to below 5.5, the processed form of toxin spontaneously inserts into the endosomal membrane; (v) once embedded in the membrane, the lipid associating regions form a pore; (vi) a proteolytic cleavage in 1.., between Fragment A and B, occurs; (vii) thereafter, Fragments A, or a polypeptide containing Fragment A, is released into the cytosol; (viii) the catalytic activity of Fragment A, i.e., the nicotinamide adenine dinucleotide-dependent adenosine diphosphate ribosylation of Elongation Factor 2, causes the death of the intoxicated cell. It is apparent that a single molecule of Fragment A introduced into the cytosol is sufficient to kill a cell.

The structural gene for diphtheria toxin, tox, is carried on a 3,900 base pair (bp) BamHl fragment

(Fig. 3) of the Corynebacteriophage β genome (Costa et al., 1981, J. Bacteriol. 148:153; Buck et al., 1981, J. Bacteriol. 148:153). This DNA fragment has three Sau3Al restriction sties which divide the tox structural gene into two major segments: the Sau3Al-2 segment has been shown to encode the diphtheria tox promoter, signal sequence, and all of fragment A (Leong et al., 1983, J. Biol. chem. 258:15016); the Sau3Al-l segment encodes all of fragment B of the toxin gene, except for the C-terminal 17 amino acids (Kaczorek et al., 1983,

Science 221:855; Ratti ete al., 1983, Nucl. Acids Res. 11:6589; Greenfield et al., 1983, Proc. Nat. Aca. sci. 80:6853). Modified Tox Gene Referring to Fig. 7, there is shown the region of plasmid pABC1508 which encodes a peptide capable of forming a chimeric toxin of the invention.

The DNA region of pABcl508 shown in Fig. 7 includes the lambda P_R promoter (substituted for the promoter naturally associated with the tox gene); an ATG initiation site; a DNA sequence encoding enzymically active Fragment A of diphtheria toxin; a portion of the DNA region encoding Fragment B of diphtheria toxin; and a linker containing a Cys codon.

Referring to Fig. 2, the portion of the diphtheria tox gene used to make a DNA sequence of the invention includes the region encoding enzymically active Fragment A ("A" in Fig. 2), and a portion of the Fragment B-encoding region ("B") at least as long as that necessary to encode the first disulfide loop ("DSL 1,"). As shown in Fig. 3, the Fragment A-encoding region (including the leader sequence) begins just downstream from a convenient Sau3AI site. The tox portion of the chimeric gene, in addition to including DSL 1-_j, may also include DSL 1_. Thus, the tox-encoding DNA sequence can end anywhere between the position just downstream of the region encoding DSL 1, and the position just downstream from the region encoding DSL 1₂; examples of such positions are, referring to Fig. 3, the Clal site, the Mspl site, the position of the Nrul site, and the SphI site. Portions of Fragment B-coding DNA located downstream of the SphI site should not be used, to avoid including the diphtheria toxin receptor binding domain. (The Nrul site in Fig. 3 ("Nrul") is not found on the wild-type

228 tox allele, but only on the mutant tox allele, described in Kaczorek et al. (1983) Science 221, 855.) Because the chimeric toxins of the invention will enter the cell via the CD4-gpl20 interaction, some or all of the pore-forming lipid associating region encoded by the region of the Fragment B-encoding DNA between Clal and Mspl can be excluded; all that is required is that all of enzymatically active Fragment A and, preferably, disulfide loop 1, are present. The Mspl site is the approximate location of the end of the region of the tox gene which encodes cross reacting material 45 (CRM 45), described in Bacha et al., id. This portion of the diphtheria toxin molecule contains the lipid associating regions of Fragment B, but does not contain 1_, and is represented in Fig. 2 as the portion of Fragment B between "y" and "z". If a DNA fragment ending at the SphI site is used, DSL1₂ is included in the encoded protein, and the portion of Fragment B is that between "y" and "x" in Fig. 2.

In the illustrated DNA construct (Fig. 7), a non-naturally occurring Cys codon, encoding a Cys residue which can be used for disulfide linkage to the CD4 analog, if a fused gene is not used, is located at the C-terminal end of the tox-encoding DNA sequence; this location ensures that the linker containing the Cys codon will not interfere with the enzymatic activity of Fragment A. Other locations in the molecule which are downstream from the Fragment A-encoding region can also be used, i.e., a Cys codon-containing linker can be inserted anywhere in the portion of the Fragment B-encoding region used. As is described below, where the chimeric toxin is formed by a disulfide linkage between the diphtheria toxin portion and the CD4 portion, the cysteine residue encoded by this introduced Cys codon is the site of that linkage; when the chimera is entirely encoded by a fused gene, the introduced Cys codon is irrelevant, as is described below. Gene Construction; pABC508

Generally, plasmids are manipulated according to standard techniques. Plasmid DNA is digested with restriction endonucleases as recommended by the manufacturer (e.g., New England Biolabs, Beverly, Mass.). Restriction fragments are electrophoresed in 1% horizontal agarose gels for 30-60 minutes at 80-100 V in TBE (89 mM boric acid, 89 mM Trizma base [Sigma Chemical Co., St. Louis, Mo.], 2.5 mM EDTA, pH 8.0) in the presence of 200 ng/ml ethidium bromide. Small DNA fragments are electrophoresed in 8% vertical polyacrylamide gels at 100 V for 2-5 hours, and stained with ethidium bromide. Gels are photographed on an ultraviolet transilluminator on Polaroid type 667 film using a red filter.

Plasmid pABC508 was constructed by fusing two pieces of DNA, one encoding Fragment A, and the other encoding part of Fragment B, to which a Cys codon-containing linker had been attached. Referring to Fig. 5, this fusion was constructed from two plasmids, pDT201, which contains the fragment A-encoding region, and pDT301, which contains most of the fragment B-encoding region of the diphtheria toxin gene. The construction of each of these pieces of DNA is described below.

Plasmid pDT301 was constructed by cutting out of the tox allele a Sau3AI-l sequence encoding all but the C-terminal 17 amino acids of Fragment B. This sequence, which carries the restriction endonuclease sites Clal, Mspl, and SphI, was inserted into the BamHI site of plasmid pUC8 (described in Viera et al. (1982) Gene 1^, 259) to yield pDT301. Plasmid pDT201 contains the Fragment A-encoding Sau3AI-2 sequence (Fig. 5) (see Leong et al. (1983) Science 220, 515). (pDT301 and pDT201, in E. coli, have been deposited in the American Type Culture Collection, Rockville, MD and given ATCC Accession Nos., respectively, 39360 and 39359. Applicant's licensee, Seragen, Inc., acknowledges its responsibility to replace these cultures should they die before the end of the term of a patent issued hereon, and its responsibility to notify the ATCC of the issuance of such a patent, at which time the deposits will be made available to the public for a period of at least 30 years after the date of deposit, Until that time the deposits will be made available to the

Commissioner of Patents under the terms of 37 CFR §1.14 and 35 USC §112.)

Still referring to Fig. 5, plasmid pDT301 was modified by the addition of a Cys codon-containing linker as follows. A synthetic linker was constructed on a controlled pore glass solid phase support in a 380A

DNA Synthesizer (Applied Biosystems, Inc., Foster City,

CA) by hybridization of 21-mer and 29-mer oligonucleotides through a 21 bp homologous core, leaving a 4bp 1/2 SphI and 1/2 Hindlll^" single-stranded sequence on each end. This linker has the sequence

AlaAlaAlaCysStp 5 '-CGGCTGCAGCATGTTAGTAGA-3 ' 3 '-GTACGCCGACGTCGTACAATCATCTTCGA-5' 1/2 SphI PstI 1/2 HindiII

The linker encodes three alanine residues, and contains a Cys codon (TGT) and a Stop codon (TAG) . (This design not only allows for the expression of a Cys-containing peptide according to the present invention, but also could allow for the insertion at the PstI site of a DNA sequence encoding a CD4 analog.)

As shown in Fig. 5, ρDT301 was digested with SphI and HindiII to remove the DNA region designed "E" in Fig. 3, and the Cys codon-containing linker was then ligated into the plasmid at the SphI, HindiII sites to give plasmid pBC508. pBC508 was then cut with HindiII and Sau3AI to give Fragment l . Still referring to Fig. 5, plasmid pDT201 was digested with HindiII and the single-stranded ends filled in with DNA polymerase I (Klenow fragment). The resulting blunt ends were ligated to the double-stranded EcoRI linkers

5'CCTTAAGG3' 3'GGAATTCC5'

(NewEngland Biolabs, Beverly, A) to give pDT201', which was then cut with EcoRI and Sau3A to give Fragment 2.

Fragments 1 and 2 were mixed in equimolar concentrations, ligated together, according to standard procedures, and the mixture was then digested with EcoRI and Hindi11.' The digested mixture was then ligated into the EcoRI and HindiII digested pEMBL8 (Dente et al. (1983) Nucleic Acid Res. Yl , 1645), which contains unique EcoRI and HindiII sites, to give pABC508. Plasmid pABC508 can be transformed into a suitable host, e.g., E. coli, as described below, to produce Cys-modified toxin molecules.

Alternatively, the naturally occurring tox promoter can be replaced with a different promoter, as follows.

The lambda P_R promoter is contained in the expression vector pEMBL8ex3 (Dente et al., id) . Referring to Fig. 6, the DNA sequence around the initiation site of the tox gene is shown, as are the corresponding amino acids. pABC508 was cut with EcoRI and then treated with Bal31 for a period of 10-15 minutes at 37°C with one unit of enzyme per microgram of DNA. The resulting mixture of DNA fragments was ligated to the BamHI linkers

5'CCTAGGCC3' 3'GGATCCGG5' (Bethesda Laboratories, Gaithersburg, MD), transformed into E. coli HB101 and the DNA sequence of the region encoding the 5' end of tox, and the sequence of 30 of the resulting clones determined. One clone, containing the DNA sequence shown in Fig. 6, was purified and the BamHI-Hindlll fragment isolated and inserted into pEMBL8ex3 which had been cut with BamHI and Hindlll. The resulting plasmid, pABC1508, contains the lambda P„ promoter and an ATG translational start codon. An extra asparagine and proline residue are inserted during this process. In Fig. 6, Cro represents the Cro gene of lambda, and SD represents the Shine-Dalgarno sequence. The lambda P„ promoter can be regulated by the lambda cl gene. In this example the mutant cl_g57 temperature-sensitive repressor gene is used such that the P_R promoter is inactive at 30°C and active at 37°C. pABC1508 (Fig. 7) was transformed, using conventional techniques (e.g., as described in Maniatis et al. (1984) Molecular Cloning: A Laboratory Manual, cold Spring Harbor, N.Y.), into E. coli HB101 (others, e.g., E. coli JM101 or SY327, can also be used) and the expression of the diphtheria tox gene products analyzed. The introduction of the positively charged asparagine residue in the tox signal sequence does not affect the export of the tox polypeptides into the periplasmic compartment of the recombinant host. E. coli cells transformed with vectors containing Cys-modified toxin-excoding DNA are grown under standard culture conditions, e.g., in Luria Broth containing, per liter, 10 g tryptone, 10 g NaCl, and 5 g yeast extract, and supplemented with 100 μg/ml ampicillin. The diphtheria toxin-related molecules, which are exported to the periplasmic space, are purified from periplasmic extracts. Periplasmic extracts are prepared from cells grown in 9.5 liter volumes at 37°C to an A__go of approximately 1.0. If the natural tox promoter has been replaced with temperature sensitive CI857 regulatory sequences under the control of the temperature-sensitive cl_₅₇ gene, as described herein, cells are grown at 30°C, and expression is induced by increasing the incubation temperature to 42°C for 15 min. The culture is then grown at 40°C for an additional hour. In either instance, the culture is concentrated to approximately 1 liter by filtration through 0.45μ membranes (Pellicon system, Millipore Corp., Bedford, Mass.) and chilled to 4°C. Bacteria are harvested by centrifugation, resuspended in ice cold 20% sucrose, 30mM Tris-HCl, 1 mM EDTA, pH 7.5, and then digested with lysozyme (750 μg/ml final concentration) for 30 minutes. Spheroplasts are removed by centrifugation, 2 mg p-amidinophenylmethylsulfonylfluoride (p-APMSF, Calbiochem, San Diego, Calif.) is added, and the periplasmic extract is sterilized by filtration through 0.2μ membranes.

The Cys-modified toxin-related molecules are then purified by chromatography on Phenyl-Sepharose (Pharmacio Fine Chemicals, Piscataway, N.J.) and

DEAE-cellulose essentially as described by Rappuoli et al. (1985) Biotechnology, p. 165. Periplasmic extracts are dialysed against lOmM sodium phosphate (pH 7.2) buffer, and ammonium sulfate added to 13% (w/v) . The crude extracts are then applied to a Phenyl-Sepharose column equilibrated with 10 mM phosphate buffer containing 13% ammonium sulfate. The modified toxin is eluted and dialysed against 10 mM phosphate buffer, and then applied to DEAE-cellulose column. After washing with phosphate buffer, the DEAE-cellulose column is developed with a linear NaCl gradient in phosphate buffer.

The modified toxin is then applied to an anti-diphtheria toxin immunoaffinity column, containing antibody made as described in Zucker et al. (1984) Molecular Immunol. 21, 785. Following extensive washing, the modified toxin is eluted with 4 M guanidine hydrochloride, and immediately dialysed against phosphate buffer. The purified modified toxin is then concentrated to approximately 100 μg/ml by placing the dialysis bag in dry Sephadex G-200. All purification procedures are carried out at 4°C, and the modified toxin is stored in small aliquots at -76°C until used. Addition of a Cys Codon to CD4 Fragments

The first step is the manipulation of CD4-encoding DNA to provide a region coding for a soluble CD4 fragment (the entire CD4 molecule is insoluble) . This is carried out as described in the literature (Hussey et al., 1988, Nature 331:78-81; Fisher et al., 1988, Nature 331:76-78; Deen et al., 1988, Nature 331:82-84; Traunecker et al. , 1988, Nature 331:84-86), as summarized below.

Plasmid pSP65-T4 (Hussey et al., id) contains the entire cDNA coding region for the CD4 receptor on a BamHI-XhoI insert (Fig. 8). The CD4 insert is isolated by digestion of the plasmid with BamHI and Xhol.

A soluble CD4 fragment can be made by deleting the sequences encoding the hydrophobic transmembrane region of CD4 (Fig. 1, designated "Tm"). The absence of this hydrophobic region renders CD4 soluble by virtue of its being incapable of hydrophobic interaction with the cellular membrane. The restriction endonuclease Neil cleaves the CD4 encoding fragment at three sites: nucleotides 83, 1253, and 1604 of the CD4 structural gene (Fig. 8). The BamHI-XhoI fragment isolated as described above can then be partially cleaved with Neil, and the fragment of 1336 base pairs, corresponding to the amino terminal encoding portion of the CD4 up to nucleotide 1253, isolated according to conventional techniques . This 1336 bp fragment encodes the first 445 amino acids of the CD4 receptor; it lacks the carboxy terminal region, including the transmembrane binding portion of the molecule.

If desired, a Cys-containing linker can at this point be fused to or near the N-terminal end of the truncated CD4 gene, in a manner analogous to that described above for the tox gene, to produce a modified CD4 fragment containing an N-terminal Cys capable of reacting with the added Cys of the diphtheria toxin fragment. Alternatively, the Cys-containing CD4 fragment can be chemically linked to a toxin portion on which a reactive sulfhydryl group has been added post-translationally, i.e., at the protein chemistry, not the DNA, level. Chemical Linkage After an available Cys has been added to both portions of the chimera, the two portions are coupled by reducing both compounds and mixing them, in a toxi :CD4 ratio of about 1:5 to 1:20; the disulfide reaction is allowed to proceed at room temperature to completion (generally, 20 to 30 minutes). The mixture is then dialyzed extensively against phosphate buffered saline to remove unreacted molecules. The final purification step involves the separation, on the basis of size, of the desired chimeric conjugates from toxin-toxin and CD4-CD4 dimers; this is done by carrying out, in phosphate-buffered saline, Sephadex G100 chromatography. Fused Gene Construction

The strategy for the genetic construction of diphtheria toxin-CD4 chimeric gene is schematically outlined in Fig. 8. The strategy involves insertion of CD4 encoding DNA into an appropriate site, for example, the Clal site of the diphtheria toxin Fragment A-encoding DNA. Plasmid pABC508 (Bishai et al. , 1987, J. Bacteriol. 169:1554; Fig. 7 hereof), which carries the diphtheria toxin promoter and structural gene up to and including the Ala._g5 codon, is digested with Clal and HindiII and treated with calf intestine alkaline phosphatase (CIAP), and ligated to the CD4-encoding DNA fragment from plasmid pSP65-T4.

Referring to Fig. 8, and the plasmids described above, the 1336 bp BamHl-Ncil CD4 fragment from pSP65-T4 is ligated to Clal + HindiII digested pABC508 in the presence of a BamHI-Clal oligonucleotide linker, which has the following nucleotide sequence:

5'-GATCCCGGGCCAT-3'

3'-GGCCCGGTAGC-5' l/2BamHl l/2ClaI Each strand of the oligonucleotide.can be synthesized using standard beta-cyanoethyl phosphoramidite chemistry, and then annealed and phosphorylated, to generate a double stranded synthetic linker molecule. The linker will form a bridge between the BamHI cohesive end of the CD4 DNA fragment and the Clal cohesive end of pABC508 plasmid. The Neil end of the CD4 fragment and the HindiII end of pABC508 can then be joined by filling in the ends with Klenow fragment of DNA polymerase I to form blunt ends, and then blunt-end ligating the filled-in ends. The chimeric toxin-CD4 fusion gene is thus created in the proper translational reading frame at the toxin Clal site, so that the first amino acid encoded by the CD4 DNA fragment is joined upstream to the truncated form of the toxin structural gene.

Smaller portions of CD4, e.g., 10-50 amino acids in length, can be made by DNA synthesized in vitro, according to procedures well-known to those skilled in the art. Following synthesis, purification and hybridization, the synthetic soluble CD4 encoding

DNA can be cloned into the Clal site of plasmid pABC508 and its sequence verified by the dideoxy chain termination method of Sanger et al. , 1977, Proc. Nat. Aca. Sci. 74:5463. Production of Chimeric Toxin Protein

Cloned diphtheria toxin gene products possessing a functional signal sequence are expressed and exported to the periplasmic compartment of E. coli K-12. The chimeric toxin can be produced using E. coli strain SY327 (Isberg et al., 1982, Cell 30:883).

Chimeric toxin-CD4 encoding DNA can be transformed into this strain according to conventional techniques. The transformed strain can then be grown in 10 liter volumes of Luria broth (10 g tryptone, 10 g NaCl, 5 g yeast extract) at 30°C for 15-18 hours. Bacteria are then harvested by centrifugation, washed, and resuspended in TES buffer (50 mM Tris-HCl, 1 mM EDTA, 20% sucrose, pH 8.3) at 4 C. Periplasmic extracts are prepared as decribed by Leong et al. (1983, Science 220:515). Bacteria can be treated with lysozme (500 μg/ml) for 20 min. and the spheroplasts sedimented by centrifugation at 10000 g for 15 min. The supernatant fluid (crude periplasmic extract) is then concentrated 4- to 6- fold by membrane ultrafiltration (YM-10, Amicon, Inc., Lexington, MA) and applied to an anti-CD4 immunoaffinity matrix (Celltech Ltd., Berkshire, UK). Following extensive washing, CD4-toxin can be eluted with 4 M guanidine hydrochloride in 0.1 M Tris-HCl, ph 9.0. Chimeric toxin preparations are then exhaustively dialysed and stored at ^~76 C until used. Purified CD4-toxin can be quantitated by absorbance at 280 nm. Extracts can be analysed for the presence of the chimeric toxin-CD4 protein by polyacrylamide gel electrophoresis, according to conventional procedures, and immunoblotting with antibodies specific for either CD4 (Becton Dickson & Co., Mountain View, CA) or diphtheria toxin (Connaught Laboratories, Toronto, Ontario, Canada), as described by Towbin et al. (1979, Proc. Nat. Aca. Sci. 476:4350); and Robb et al. (1982, Infect. Immunol. 38:267). Mechansim of Action

A chimeric toxin of the invention will be bifunctional; that is, the CD4 portion will retain the ability of the CD4 receptor to specifically recognize and bind to the gpl20 envelop protein of HIV, and the Fragment A diphtheria toxin portion will retain toxicity. The mode of entry into eukaryotic cells is by association of the chimeric toxin with potentially infecting virions or with gpl20 present on HIV-infected cells. That is, once the chimeric toxin is administered to a human patient infected with HIV, the toxin will target either freely circulating or budding virions and become associated with them by virtue of noncovalent interaction between the CD4 portion of the chimeric toxin and the gpl20 envelop protein of the virus. The virus, or the toxin will target gpl20 expressed on the surface of already infected cells. In the former instance, the virus becomes a carrier for entry of the chimeric toxin into a CD4 cell via the CD4 receptor; in the latter case, CD4 is itself the carrier, and gpl20 the target. In either case, once the chimeric toxin enters the cell, the toxin portion of the chimeric protein arrests cellular functions. Only potentially infectible CD4 cells (primarily T-lymphocytes and macrophages) are killed, and thus potential reservoirs for HIV are eliminated without harming CD4~ cells.

Use The chimeric toxin can be admixed with a pharmaceutically acceptable carrier substance, e.g., saline, and administered, by a medically acceptable administration route, e.g., intravenously, orally, or intramuscularly, to patients infected with HIV, to destroy CD4 cells that are infected, or potentially infectible, with HIV. The amount of chimeric toxin administered will generally be between 100 μg/kg body weight and 1 mg/kg body weight per day.

Other Embodiments Other embodiments are within the following claims. For example, any soluble gpl20-binding CD4 fragment can be used as the CD4 portion of the chimeric toxin; it is only required that all or most of the transmembrane portion of CD4 be excluded, so that the fragment is soluble, and that the fragment be sufficiently long and sufficiently homologous to the natural molecule to bind to gpl20. Fragments can be generated at the DNA level using conventional recombinant DNA techniques or at the protein level, using proteolysis. A small portion, i.e., six or fewer amino acids, of the transmembrane portion of CD4 can be included without preventing solubility. Soluble CD4 fragments can be tested to ensure that they have retained the ability to selectively bind gpl20 as follows. Metabolically labeled gpl20 can be incubated with unlabelled purified soluble CD4 either in the absence or presence of monoclonal antibodies directed against distinct epitopes of the CD4 protein (OKT4 and OKT4A) . The OKT4A antibody is known to inhibit the binding of gpl20 to the CD4 molecule. If soluble CD4 is capable of binding gpl20, then preincubation of CD4 with OKT4A antibody will inhibit gpl20 co-precipitation. These procedures are performed according to techniques well known to those skilled in the art. Although it is not known what region of the CD4 molecule binds gpl20, nor whether the same or different regions bind to an invariant region of class II MHC molecule(s) which are the presumed physiologic CD4 ligand, sequence analysis of CD4 has suggested an evolutionary origin from a structure with four immunoglobulin-related domains. Two of these domains (I and II) are involved in gpl20 binding.

Claims

Claims 1. A chimeric toxin composed of a first portion comprising CD4, or a gp 120-binding analog or portion thereof, covalently bonded to a second portion comprising a portion of diphtheria toxin capable of killing a human T-lymphocyte and incapable of causing generalized binding of said chimeric toxin to human cells.

. 2. The toxin of claim 1, said toxin being water soluble.

3. The toxin of claim 2 wherein said portion or analog of CD4 does not include the transmembrane region of CD4.

4. The toxin of claim 1 wherein said first and second portions are bonded to each other via a peptide bond.

5. The toxin of claim 4 wherein said toxin is encoded by a fused gene including regions encoding said first and second portions.

6. The toxin of claim 5 wherein, in said fused gene, the coding region encoding said second portion is located upstream of and in reading frame with the coding region encoding said first portion.

7. The toxin of claim 1 wherein said second portion includes the enzymatically active Fragment A of diphtheria toxin and the disulfide loop cleavage domain adjacent said Fragment A of diphtheria toxin. 8. The toxin of claim 1 wherein said first and second portions are linked by a non-peptide covalent linkage.

9. The toxin of claim 8 wherein said linkage is a disulfide linkage between a cysteine residue on said first portion and a cysteine residue on said second portion.

. 10. The toxin of claim 9 wherein said cysteine residue on said first portion is encoded by a cysteine codon which does not occur naturally in the gene encoding CD4, and said cysteine residue on said second portion is encoded by a cysteine codon which does not occur naturally in the gene encoding diphtheria toxin.

11. The toxin of claim 1, said toxin being complexed with a gpl20 molecule.

12. The toxin of claim 11, wherein said gpl20 molecule is present on the surface of an HIV virion.

13. The toxin of claim 12, wherein said virion is in the bloodstream of a human patient.

14. The toxin of claim 11, wherein said gpl20 molecule is present on the surfact of an HIV-infected cell.

15. A method of killing CD4⁺ cells of a human patient infected with HIV, comprising administering to said patient a cell-lysing amount of the toxin of claim 1. 16. The method of claim 15 wherein said toxin is capable of complexing with cell-free HIV in the blood of said patient, said HIV having surface gpl20 molecules capable of targeting said toxin to said CD4⁺ cells.