WO1991019745A1

WO1991019745A1 - Chimeric toxins with improved inter-domain geometry

Info

Publication number: WO1991019745A1
Application number: PCT/US1991/004187
Authority: WO
Inventors: John R. Murphy; Diane P. Williams
Original assignee: The University Hospital
Priority date: 1990-06-13
Filing date: 1991-06-12
Publication date: 1991-12-26
Also published as: JPH06503553A; CA2083487A1; AU8056691A; EP0535088A1; EP0535088A4

Abstract

A chimeric toxin including protein fragments joined together by peptide bonds, the chimeric toxin comprising, in sequential order, beginning at the amino terminal end of the chimeric toxin: (a) the enzymatically active Fragment A of diphtheria toxin; (b) a first fragment including the cleavage domain 11 adjacent Fragment A of diphtheria toxin; (c) a second fragment comprising at least a portion of the hydrophobic transmembrane region of Fragment B of diphtheria toxin, the second fragment having a deletion of at least 50 diphtheria toxin amino acid residues, the deletion being C-terminal to the portion of the transmembrane region, and the second fragment not including domain 12; (d) a spacer; (e) a portion of a cell-specific polypeptide ligand, the cell-specific polypeptide ligand being a cell growth factor, the portion including at least a portion of the binding domain of the polypeptide ligand, the portion of the binding domain being effective to cause the chimeric toxin to bind selectively to the target cell.

Description

_ -_ _

CHIMERIC TOXINS WITH IMPROVED INTER-DOMAIN GEOMETRY

Background of the Invention This invention relates to the use of recombinant DNA techniques to construct chimeric toxin molecules. The highly selective effects asserted by many hormones, toxins, and other biologically active proteins are in part possible because such proteins possess more than one functionally distinct polypeptide domain. Some plant and bacterial toxins, e.g., have evolved with separate domains responsible for cell binding, membrane translocation, and intoxication. The combination of properties conferred by the various domains results in extremely potent bioactive molecules.

The diphtheria toxin (DT) is an example of a naturally occurring multi-domain protein. DT consists of a number of domains, each of which confers a particular function, and all of which, in combination, result in an extraordinarily active toxin molecule. DT can be characterized, starting at the amino terminal end of the molecule, as follows: a hydrophobic leader signal sequence s (amino acids Val_₂₅ - Ala ,); enzymatically-active Fragment A (amino acids Gly. - Arg_ιg3) which includes a domain which catalyzes the the nicotinamide adenine dinucleotide-dependent adenosine diphosphate (ADP) ribosylation of the eu aryotic protein synthesis factor termed "Elongation Factor 2"; the protease-sensitive disulfide loop 1. (amino acids Cys₁₈₆ - Cys₂₀₁), which contains a cleavage domain; and Fragment B (amino acids Ser-_g. - Ser₅₃₅), which includes a translocation domain and a generalized binding domain flanking a second disulfide loop d₂' amino acids Cys.g. - Cys.__) . The process by which DT intoxicates sensitive eukaryotic cells involves at least the following steps: (1) the binding domain of diphtheria toxin binds to specific receptors on the surface of a sensitive cell; (ii) while bound to its receptor, the toxin molecule is internalized into an endocytotic vesicle; (iii) either prior to internalization, or within the endocytotic vesicle, the toxin molecule undergoes a proteolytic cleavage in the 1. cleavage domain between Fragments A and B; (iv) as the pH of the endocytotic vesicle decreases to below 6, the toxin spontaneously inserts into the endosomal membrane; (v) once embedded in the membrane, the translocation domain of the toxin facilitates the delivery of Fragment A into the cytosol; (vi) the catalytic activity, which resides in a domain in Fragment A, causes the death of the intoxicated cell. The mechanism of cell killing by Pseudomonas exotoxin A, and possibly by certain other naturally-occurring toxins, is very similar.

The therapeutic potential of man-made toxin molecules containing various functional domains (whether polypeptide or nonpeptide) has been appreciated for many years. Paul Ehrlich, Ehrlich (1906) in Collected Studies on Immunity 2:442-447, was the first to suggest the construction of bifunctional molecules that combined a molecule with an affinity for a specific target (a targeting or cell-binding domain) with a molecule that acts as a cytotoxic agent (a cytotoxic domain) . Since that time numerous biological and chemical moieties have been coupled in a variety of ways in attempts to create molecules that exhibit a degree of selective, i.e., targeted, cytotoxicity. Early efforts were typified by the conjugation of non-peptide toxins to solubilizing molecules or to antibodies. More recently, the advent of molecular genetics has allowed the engineering of chimeric genes that encode novel multi-domain chimeric toxin proteins. See, e.g., Murphy U.S. Patent No. 4,675,382, hereby incorporated by reference, which teaches the construction of hybrid proteins that include the enzymatically active (toxin) domain of DT, the cleavage domain of DT, the translocation domain of DT, and a cell-specific binding domain derived from a second protein. The hybrid proteins described in Murphy supra combine DT toxin domains, which confer toxicity, with a domain from a different protein, e.g., interleukin 2 (IL-2), which confers extremely selective cell binding properties. Improvements in the intrinsic properties of the constituent components, i.e., the use of more highly-specific cell-binding agents, e.g., monoclonal antibodies, and the use of toxins of increased potency, e.g., plant or bacterial toxins, have been the primary routes to improved toxin conjugates. The way in which the cell binding and cell-killing entities of bifunctional molecule are coupled has also received attention in attempts to improve the performance of these molecules. These efforts have been directed to preserving the activities of the primary components, increasing solubility, increasing the ratio of cytotoxic agent bound to cell specific binding agent, or increasing the ease with which a required step in intoxication, e.g., the cleavage of a toxin domain from the rest of the molecule, is carried out.

Various spacer or linker members have been placed between the domains (or nonpeptide functional entities) of conjugate-toxins in efforts to realize the advantages discussed above. Rowland et al. (1975) Nature 255:487 reports that the use of polyglutamic acid

(molecular weight=35,000) as a linker between chemical toxins and antibodies was preferable to direct linkage of the toxin to the antibody.

"By minimizing interference with the chemical structure of the Ig in this linkage step, a conjugate is produced with both a high concentration of drug and little loss of antibody activity . . . The concept of linking cytotoxic drugs to antibody through an inert intermediate carrier offers a wide scope for improved cancer chemotherapy in the future, with the possibility of using a variety of drugs, different carriers and antibody preparations of greater purity."

Monsigny et al. (1980) FEBS Letters 119:181 reports maximizing the activity of a drug, e.g. , daunorubicin conjugated to a carrier, e.g., an antibody, by the insertion of a peptide containing spacer between the drug and the carrier. The spacer arm, 2-(1-thio-β-D-glycopyranosyl)-ethanoyl-L-arg-L-leu , can be cleaved by lysosomal but not by serum proteases. "For technical reasons, and because the activity of a drug is partially or totally lost when it is substituted or chemically modified, we devised a spacer arm such that the drug—carrier conjugate is stable in serum and can be specifically split by lysosomal proteases leading to the free drug inside the target cells."

Arnon et al. (1982) Immunol Rev. 62:5 reports the use of dext an to link daunomycin to an antibody.

"The reason for employing this procedure was two-fold. First, it is expected to result in higher extent of drug binding per antibody molecule, which should lead to higher cytotoxic activity; second, the use of a macromolecule as a spacer arm between the drug and the carrier could prove advantageous in permitting higher exposure of the drug moiety on the conjugate surface with less steric * hindrance and hence higher efficacy."

Truet et al. (1982) Proc. Natl. Acad. Sci. USA, 79.:626-629 reports coupling daunorubicin to succinylated bovine serum albumin (BSA) by a spacer arm one to four amino acid residues in length. The purpose of the spacer arm was to allow lysosomal cleavage between the drug and the BSA molecule.

Neville et al. (1989) JBC 264:14653-14661 reports that a cleavable cross-linker enhances potency of a DT-antigen conjugate three to ten fold. The crosslinkers are cleavable at acid pH, and are thus cleaved in an acidic compartment. The increased potency is believed to be due to an enhanced intracellular toxin-toxin receptor interaction which leads to increased translocation; the conjugate is thought to be sterically hindered prior to, but not after, cleavage. A nonpeptide crosslinker, bis-maleimidoethyoxy propane, was used.

Greenfield et al. PCT/US85/00197 describes a toxin-antibody conjugate wherein the toxin and antibody are coupled by means of a spacer peptide.

"the invention is designed to provide a toxin conjugate which has the appropriate geometry for translocating the cytotoxic fragment into the target cell, the capacity to retain its binding fragment prior to such translocation, and/or the ability to solubilize the cytotoxic portion. In one aspect of the invention, the spacer is designed so as to permit the cytotoxic portion of the molecule ready access to the cell membrane. As the size of a typical antibody (binding) fragment is very much greater than that of most cytotoxic. fragments, there is considerable steric hinderance of the access to the cell membrane by the cytotoxic portion imposed by the sheer bulk of the antibody or antibody fragments . . . In order to effect this translocation, the spacer needs to be sufficiently flexible to allow the A portion to reach the cell membrane, and sufficiently extended to permit it to have sufficient reach."

Summary of the Invention In general, the invention features a chimeric toxin including protein fragments joined together by peptide bonds, including, in sequential order, beginning at the amino terminal end of the chimeric toxin: (a) the enzymatically active Fragment A of diphtheria toxin; (b) a fragment including the cleavage domain 1. adjacent Fragment A of diphtheria toxin; (c) a fragment including (i) at least a portion of the hydrophobic transmembrane region of Fragment B of diphtheria toxin, the fragment having a deletion of at least 50, preferably of at least 80, diphtheria toxin amino acid residues, the deletion being C-terminal to the portion of the transmembrane region, and the fragment not including domain 1₂, or, (ii) a fragment including at least a portion of the hydrophobic transmembrane region of Fragment B of diphtheria toxin wherein said Fragment B of diphtheria toxin does not include any diphtheria toxin sequences C-terminal to amino acid residue 386 of native diphtheria toxin; (d) a spacer (defined infra); and (e) a portion of a cell-specific polypeptide ligand, the cell-specific polypeptide ligand being a cell growth factor preferably a lymphokine, e.g., interleukin 2 (IL-2), or interleukin 4 (IL-4). The portion of the cell-specific polypeptide ligand includes at least a portion of the binding domain of the polypeptide ligand, the portion of the binding domain being effective to cause the chimeric toxin to bind selectively to the target cell e.g., lymphocytes, e.g., T-cells or B-cells bearing receptors for the ligand. Cell growth factor, as used herein, means a protein that binds to a cell surface receptor found on a mammalian cell and causes proliferation of the cell. Preferred cell growth factors are lymphokines, i.e., cell growth factors that bind to and stimulate the proliferation of lymphocytes.

In preferred embodiments DAB_3gg is the cytotoxic portion, i.e., (a), (b), and (c) above. (DAB_38g consists of methionine followed by residues 1-386 of native DT followed by residues 484 and 485 of native DT.) The construction of DAB_38g is discussed in USSN 488,608, filed March 2, 1990, hereby incorporated by reference.

Preferred embodiments include those in which the spacer: is at least 5 amino acids in length, preferably 10-30 amino acids in length; when placed between the sequence of DT fragment DAB.₈₅ (DAB_4g5 consists of methionine followed by the first 484 amino acid residues of native DT) and amino acid residues

2-133 of IL-2, has a Bn„_orm value of 1.000 or g ^~reater, more preferably of 1.125 or greater, and most preferably of 1.135 or greater; is composed of at least 60%, and preferably of at least 80% of amino acids from the group of lysine, serine, glycine, proline, aspartic acid, glutamic acid, glutamine, threonine, asparagine, or arginine; is at least 60%, and preferably at least 80%, homologous to any of

Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr, hereinafter referred to as (1-10), Pro-Lys-Ser-Gly-Thr-Gln-Gly, hereinafter referred to as (1-7 Gly), Pro-Thr-Ser-Ser-Ser-Thr-Lys, (hereinafter referred to as (1-7 Lys), or multiples thereof (the number of multiples hereinafter referred to with a superscript, e.g., (1-10) , which indicates 2 tandem copies of the (1-10) subunit; results in an affinity of the chimeric toxin for the target cells that is greater than the affinity of a second chimeric toxin for the target cells, the second chimeric toxin being identical to the chimeric toxin except that the second chimeric toxin lacks the spacer; or, results in the chimeric toxin exhibiting cytotoxicity for the target cells that is a least 2 times greater than the cytotoxicity exhibited by a second chimeric toxin for the target cells, the second chimeric toxin being identical to the chimeric toxin except that the second chimeric toxin lacks the spacer. Diphtheria toxin, or native diptheria toxin, as used herein, means the 535 amino acid residue mature form of diphtheria toxin protein secreted by Corynebacterium diphtheriae. The sequence of an allele of the gene which encodes native diphtheria toxin can be found in Greenfield et al. (1983) Proc. Natl. Acad. Sci. USA 80_:6853-6857, hereby incorporated by reference. Enzymatically active Fragment A, as used herein, means amino acid residues Gly 1 through Arg 193 of native DT, or an enzymatically active derivative or analog of the natural sequence. Cleavage domain 1-, as used herein, means the protease sensitive domain within the region spanning Cys 186 and Cys 201 of native DT. Fragment B, as used herein, means the region from Ser 194 through Ser 535 of native DT. The hydrophobic transmembrane region, or hydrophobic domain, of Fragment B, as used herein, means the amino acid sequence bearing a structural similarity to the bilayer-spanning helices of integral membrane proteins and located approximately at or derived from amino acid residue 346 through amino acid residue 371 of native diphtheria toxin. Domain 1₂, as used herein, means the region spanning Cys 461 and Cys 471 of native DT. The generalized eukaryotic binding site of Fragment B, as used herein, means a region within the C-terminal 50 amino acid residues of native DT responsible for binding DT to its native receptor on the surface of eukaryotic cells. The generalized eukaryotic binding site of Fragment B is not included in the chimeric toxins of the invention. A spacer, as used herein, is a polypeptide which possesses one or more of the following characteristics: (1) when placed between the sequence of the diphtheria toxin fragment DAB._g5 and amino acid residues 2-133 of IL-2, it possesses an amino acid residue with a normalized B value (^B _norm) (as defined in Karplus et al. (1985) Naturwissenschaften 72:212, hereby incorporated by reference) of 1.000 or greater, preferably of 1.125 or greater, and most preferably of 1.135 or greater; (2) at least 60% and preferably at least 80% of its amino acid residues are from the group of lysine, serine, glycine, proline, aspartic acid, glutamic acid, glutamine, threonine, asparagine, or arginine; (3) it possesses the sequence, or multiples of the sequence Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr; (4) it is at least 60% and preferably at least 80% homologous to the sequence, or to multiples of the sequence listed in (3); (5) it possesses the sequence, or multiples of the sequence Pro-Lys-Ser-Gly-Thr-Gln-Gly; (6) it is at least 60% and preferably at least 80% homologous to the sequence, or multiples of the sequence, listed in (5); (7) it possesses the sequence, or multiples of the sequence of Pro-Thr-Ser-Ser-Ser-Thr-Lys; or (8) it is at least 60% and preferably at least 80% homologous to the sequence, or multiples of the sequence, listed in (7).

The Bnorm value of residues in a p ^coly^Jp^JΓep ^~tide sequence can be determined with computer programs, e.g., with FLEXPRO (Intelligenetics, Mountain View (CA)) which predicts the flexibility between alpha carbon atoms at each point of a selected protein sequence, using the method (the B_norm method) of Karplus et al., supra. FLEXPRO calculates the chain flexibitiy at a selected amino acid residue from the average values of the atomic temperature factors (also called B values or

Debye-Waller factors) of the alpha carbon atoms in adjacent amino acids. The B value is the mean square displacement of the atom from its average position in the protein. Flexible locations have high B values because their displacement can be large.

The Bnorm value for an amino acid in FLEXPRO is affected by -' the Bnorm values of its neig -hboring ^~ amino acids. The predicted flexibility at an amino acid calculated by FLEXPRO is the weighted sum of the Bnorm values (taking account of neighbors) of the seven amino acids closest to that point in the sequence. The weight for the two outermost amino acids is 0.25; for the two next to them, 0.5; for the two adjacent to the central amino acid, 0.75; and for central amino acid itself, 1. For each amino acid, the weight is multiplied by the

Bnorm value (taking account of neighbors) . When an amino acid sequence is input to FLEXPRO, the program calculates the Bnorm values and disp ^~lay ^Λs 7-amino acid residue sections and the p ^~~eak B„norm value for each of the 7-amino acid residue sections.

A normalized B value of less than 1 indicates a rigid amino acid; a value greater than 1 indicates a flexible amino acid. Observations in a number of proteins suggest that the amino acids alanine, valine. leucine, isoleucine, tyrosine, phenylalanine, tryptophan, cysteine, methionine, and histidine tend to be rigid and that the amino acids lysine, serine, glycine, proline, aspartic acid, glutamic acid glutamine, threonine, asparagine, and arginine tend to be flexible. (Karplus et al. supra)

The sequence used as a spacer may be derived from any source e.g., from one of the polypeptides used to construct the chimeric toxin, from other naturally occuring sequences, or from synthetic sequences regardless of whether they are naturally occuring.

The invention also features a chimeric toxin encoded by a fused gene including regions coding for the protein fragments, a DNA sequence encoding the chimeric toxin, an expression vector containing the DNA sequence encoding the chimeric toxin, a cell transformed with the expression vector, and a method of producing the chimeric toxin including culturing the transformed cell and isolating said chimeric toxin from the cultured cell or supernatant.

The invention also features spacer peptides having the sequence

Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr, or tandem repeats of that 10 residue subunit (with or without one or two additional residues at each end (or between the individual subunits of the tandem repeat) that arise from the inclusion of functional or disfunctional 1/2 restriction enzyme recognition site linker sequences in the DNA that encodes the spacer), DNA segments encoding the spacer peptides (with or without functional or disfunctional 1/2 restriction enzyme recognition site linker sequences), vectors containing these DNA segments, cells transformed with these vectors, and methods of making the spacer peptide including culturing the transformed cell, and isolating the spacer from the cultured cell or supernatant.

Preferred embodiments include a DNA segment encoding the spacer wherein the spacer is a tandem repeat of the subunit

Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr, and the DNA sequence encoding each occurence of the subunit in the spacer is nonhomologous with the DNA sequence encoding every other occurence of the subunit in the spacer or the chimeric toxin, the nonhomology being sufficient to prevent recombination between sequences encoding tandemly repeated subunits.

The invention also features a method of preventing recombination between the tandemly repeated subunits of spacer-peptide-encoding DNA by choosing the codons of each subunit-encoding sequence such that the DNA encoding each subunit is nonhomologous with the DNA encoding every other subunit, the nonhomology being sufficient to prevent recombination between tandemly repeated subunits.

The invention also features DNA encoding a spacer (with or without functional or disfunctional 1/2 restriction enzyme recognition site linker sequences), an expression vector containing that DNA, a cell transformed with that expression vector, and a method of producing the spacer including culturing the transformed cell and isolating said spacer from the cultured cell or supernatant.

Molecules of the invention exhibit improved binding affinity and improved cytotoxicity for cells bearing the receptor to which the ligand portion of the chimeric toxin binds. In the management of autoimmune disease, allograft rejection, and other lymphocyte-dependent conditions, a chimeric toxin that employs a cell binding portion that recognizes an interleukin (or other growth factor) receptor must compete with indigenous interleukin (or other growth factor) for sites on the target cell. Thus, optimization of the early, cell-binding, step is particularly critical in chimeric toxins in which the cell binding portion recognizes a ligand such as an interleukin receptor.

Other features and advantages of the invention will be apparent from the following description of the preferred embodiments and from the claims.

Description of the Preferred Embodiments

The drawings will first be briefly described.

Drawings Fig. 1 is a diagram of the DT molecule and various fusion proteins;

Fig. 2 is a depiction of the construction of the plasmids of a preferred embodiment;

Structure and Synthesis of chimeric toxins with improved inter-domain geometry

DAB4. _o86_t.-(1-10)-IL-2 is a chimeric toxin polypeptide consisting of, in the following order: Met; amino acid residues 1 through His 484 of mature native

DT; Gly; the amino acid residues Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr (indicated as

(1-10)); His; Ala 485 of mature native DT; and amino acid residues 2 through 133 of IL-2. The DT portion of the chimeric toxin DAB_48g-(l-10)- IL-2 includes all of DT fragment A and the portion of DT fragment B extending to residue 485 of mature native DT. See Fig. la for the structure of DT. Fig. lb shows the structure of

In Fig. lb a wide bar indicates the fusion protein, narrow connecting lines represent deletions, numbers above the bars are amino acid residue numbers in the DAB nomenclature (described below), numbers below the bars correspond to the amino acid residue numbering of native DT, cross hatching indicated amphir rhic regions, darkened areas correspond to the transr nbrane region, IL-2-2-133 indicates amino acid resid^* 3 2-133 of IL-2, Ala = alanine, Asn = asparagine, Asp = spartic acid, Cys = cysteine, Gly - glycine, His = histidine, lie = isoleucine, Met = methionine, Thr = threonine, Tyr = tyrosine, and Val = valine. (The nomenclature adopted for DT-IL-2 toxins is illustrated by DAB₄₈₆-(l-10)-IL-2, where D indicates diphtheria toxin, A and B indicate wild type sequences for these DT fragments, the number in the parenthesis represents a spacer polypeptide, and IL-2 indicates mature human interleukin-2 sequences. The numerical subscript indicates the number of DT-related amino acids in the fusion protein, the last of which is at the C-terminal end of the spacer where a spacer is inserted. Note that the last two codons of DT also function as a 1/2 SphI site. Since the deletion of the tox signal sequence and expression from the trc promoter results in the addition of a methionine residue to the N-terminus, the numbering of DAB-IL-2 fusion toxins is +1 out of phase with that of native diphtheria toxin.) DAB 4.8₀6_/--(1-10)-IL-2 was constructed from

DAB.₈₆-IL-2. pDW24, which carries DAB._g6-IL-2 was constructed as follows. pUC18 (New England BioLabs) was digested with PstI and Bgll and the Pstl-Bgll fragment carrying the E.coli origin of replication, the polylinker region, and the 3' portion of the β-lacatamase gene (amp^r) was recovered. Plasmid pKK-233-2 (Pharmacia) was digested with PstI and Bgll and the Pstl-Bgll fragment carrying, two transcription terminators and the 5' portion of the β-lactamase gene was recovered. pDW22 was constructed by ligating these two recovered fragments together. pDW23 was constructed by isolating a BamHI-Sall fragment encoding human IL-2 from plasmid pDW15 (Williams et al. (1988) Nucleic Acids Res. 11:10453- 10467) and ligating it to BamHI/Sall digested pDW22 (described above) . pDW24 was constructed as follows. A BamHI-Ncol fragment carrying the trc promoter and translational initiation codon (ATG) was isolated from plasmid pKK233-2 (Pharmacia). The DNA sequence encoding amino acid residues 1 through 485 of DT was obtained by digesting pABC508 (Williams et al. (1987) Protein Engineering 1.:493-498) with SphI and Haell and recovering the Haell-SphI fragment containing the sequence encoding amino acid residues 1 through 485 of DT. A Ncol/Haell linker (5'CCATGGGCGC 3') was ligated to the Haell-SphI fragment and that construction was then ligated to the previously isolated BamHI-Ncol fragment carrying the trc promoter. This results in a BamHI-SphI fragment bearing, in the following order, the trc promoter, the Ncol site (which supplies the ATG initiator codon for Met), and the sequence encoding residues 1 through 485 of native DT. This fragment was inserted into pDW23 that had been digested with Ba HI and SphI. The resulting plasmid was designated pDW24. pDW24 is shown in Fig. 2. The insert corresponding to DAB_4ββ-IL-2 is shown as a heavy line. In Fig. 2 filled circles indicate Ncol sites, open circles indicate Nsil sites, open diamonds indicate Clal sites, filled squares indicate Hpall sites, open squares indicate SphI sites, and filled triangles indicate Sail sites. The fusion protein (DAB.₈--Il-2) encoded by pDW24 is expressed from the trc promoter and consists of Met followed by amino acids 1 through 485 of mature DT fused to amino acids 2 through 133 of human IL-2.

DNA encoding the polypeptide spacer, (1-10), was synthesized and inserted into pDW24. pDW24 was cut at the SphI site at the 3' end of the DT sequence and a synthetic sequence encoding a spacer (or multiples thereof) inserted. The sequence of the synthetic sequences encoding 1, 2, and 3 copies of the (1-10) spacer are shown in Table 1.

TABLE 1: SPACER SEQUENCES

1 2 3 4 5 6 7 8 9 10

1. Ala Pro Thr Ser Ser Ser Thr Lys Lys Thr His

CT CCG ACC AGC TCT AGC ACT AAA AAG ACT CAT

GTA CGA GGC TGG TCG AGA TCG TGA TTT. TTC TGA

Functional SphI site Disfunctional SphI sit

1 2 3 4 5 6 7 8 9 10

2. Gly Ala Pro Thr Ser Ser Ser Thr Lys Lys Thr-

GT GCA CCG ACT AGC AGC TCT ACT AAG AAA ACA- GTA CCA CGT GGC TGA TCG TCG AGA TGA TTC TTT TGT-

Disfunctional SphI site

11 12 13 14 15 16 17 18 19 20

Ala Pro Thr Ser Ser Ser Thr Lys Lys Thr His GCT CCT ACC TCT TCT AGC ACG AAG AAG ACG CAT G

CGA GGA TGG AGA AGA TCG TGC TTC TTC TGC

Functional SphI site

1 2 3 4 5 6 7 8 9 10

3. Gly Ala Pro Thr Ser Ser Ser Thr Lys Lys Thr-

GT GCA CCG ACT AGC AGC TCT ACT AAG AAA ACA-

GTA CCA CGT GGC TGA TCG TCG AGA TGA TTC TTT TGT- Disfunctional SphI site

11 12 13 14 15 16 17 18 19 20 21 22

Ala Pro Thr Ser Ser Ser Thr Lys Lys Thr His Ala-

GCT CCT ACC TCT TCT AGC ACG AAG AAG ACG CAT GCT-

CGA GGA TGG AGA AGA TCG TGC TTC TTC TGC GTA CGA-

Functional SphI site

23 24 25 26 27 28 29 30 31 32

Pro Thr Ser Ser Ser Thr Lys Lys Thr His

CCG ACC AGC TCT AGC ACT AAA AAG ACT CAT G

GGC TGG TCG AGA TCG TGA TTT TTC TGA

Disfunctional SphI site A 4-base extension capable of annealing to a 1/2 SphI site is present at either end of the DNA fragment encoding each spacer sequence. In each spacer, codon substitutions destroy one of these SphI recognition sites such that only one SphI site is regenerated. Refering to Table 1, sequence 1, the use of T in place of G in the third position of the 10th codon prevents the creation of an SphI site at the 3' end of the spacer. In sequence 2, the use of G in place of C in the 2nd position of the Gly codon prevents the creation of an SphI site at the spacer's 5' end. Sequence 3 is mad by maintaining the functional SphI site at the 5' end of sequence 1 and ligating the spacer of sequence 2 at that site, leaving a regenerated SphI site 3' of the spacer of sequence 2, and disfunctional SphI sites at either end of the spacer of sequence 3.

Thus, insertion of sequence 1 from Table 1 into the SphI site of DAB 4.8_Qb.-IL-2 or DABja_Qy_O-IL-2 results₄₈in₄ a fusion protein with splice/junctions as follows: the His residue of native

DT; Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr; a His residue encoded by the 3' extension; Ala 485 of native DT; and the second amino acid of the IL-2 sequence. The insertion of sequence 2 from Table 1 into the SphI sit of DAB4.8_0/6.-IL-2 or DABJo_QQ ^y-IL-2 results₄₈in₄ a fusion protein with splice/junctions as follows: the His residue of native DT; a Gly residue derived from the 3' extension at the 5' end of the oligonucleotide; Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr- Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr; a His residue encoded by the 3' extension; Ala 485 of native DT; and the second amino acid of the IL-2 sequence.

The insertion of sequence 3 from Table 1 into the SphI sit of DAB4.8₀6.-IL-2 or DAB ά_Qa_Ov-IL-2 results₄₈in₄ a fusion protein with splice/junctions as follows: the His residue of native DT; a Gly residue derived from the 3" extension at the 5' end of the oligonucleotide;

Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr-Ala-Pro-Thr-Ser-Ser-Ser-Thr Lys-Lys-Thr-His-Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr; a His residue encoded by the 3' extension; Ala485 of native DT; and the second amino acid of the IL-2 sequence. Three subunits may be ad by ligating sequence 1 and 2 in vitro to achieve sequence 3, which is then inserted into the chimeric toxin, or, sequence 1 may be inserted into the SphI site of a chimeric toxin that harbors sequence 2.

Insertion of sequences 2 or 3 from Table 1, which encode and 3 copies of (1-10) respectively, result in analagous structures. When repeats of the (1-10) sequence are present codon are hosen such that DNA homology between sequential (1-10) encodi seqi nces is minimized. This prevents rearrangements that would arise from homologous recombination between sequential copies of (1-10) encoding DNA .

The presence of the Gly and His residues (encoded by the linkers at each end of the spacer sequence) do not affect the

Bnorm value of the linker (data not shown),

The (1-10) spacer sequence was identified by applying the FLEXPRO program (described above) to the sequence of DAB_4gg-IL-2. Table 2 shows the 10 most flexible 7-amino acid segments in DAB₄₈₆~IL-2 as determined by FLEXPRO.

TABLE 2: FLEXPRO Analysis of DAB₄₈₆-IL-2 Rank From To B_norm Sequence

1 487 493 1.135 Pro-Thr-Ser-Ser-Ser-Thr-Lys

2 39 45 1.125. Pro-Lys-Ser-Gly-Thr-Gln-Gly

3 142 148 1.107 Ala-Glu-Gly-Ser-Ser-Ser-Val

4 581 587 1.107 Leu-Lys-Gly-Ser-Glu-Thr-Thr

5 172 178 1.102 Gly-Lys-Arg-Gly-Gln-Asp-Ala

6 7 13 1.098 Val-Asp-Ser---Ser-Lys-Ser-Phe

7 232 238 1.097 Ser-Glu-Ser-Fro-Asn-Lys-Thr

8 63 74 1.036 Val-Asp-Asn-Glu-Asn-Pro-Leu

9 411 417 1.085 Phe-Gln-Gly-Glu-Ser-Gly-His 10 266 272 1.08 Thr-Val-Thr-Gly-Thr-Asn-Pro

The most flexible seven amino acid stretch of the fusion protein was found to be amino acids, residues 487 throug -h 493 of DAB4.,8₀b,.-IL-2, with a Bno_r.,m_, value of

1.135. These seven amino acids correspond to amino acid residues 2-8 of IL-2. These seven amino acids were also the most flexible found in DAB_38g-IL-2 and DAB_2g5-IL-2 (a molecule entirely devoid of the toxic characteristics of the other DT chimeric toxins tested) . Because the first 10 amino acid residues of IL-2 are known not to form an orderly array in crystals (Brandhuber (1987) Science 238:1707-1709), and are thus flexible, the entire first 10 amino acid residues of

IL-2 were used as the (1-10) spacer. When inserted into DAB₄₈₆~IL-2, the (1-10) spacer is the most flexible sequence of the new construction, as shown in Table 3.

TABLE 3: A: FLEXPRO Analysis of DAB ₈₆-(l-10)-IL-2

Rank From To B[Norm] Sequence

1 488 494 1 135 Pro-Thr- ^•Ser-Ser-Ser- ^•Thr-Lys 2 498 504 1 135 Pro-Thr- ^■Ser-Ser-Ser- ^•Thr-Lys 3 509 515 1 135 Pro-Thr- Ser-Ser-Ser- ^•Thr-Lys

4 39 45 1 125 Pro-Lys- Ser-Gly-Thr- ^•Gln-Gly 5 142 148 1 107 Ala-Glu- ^•Gly-Ser-Glu- ^■Thr-Thr 6 603 609 1 107 Leu-Lys- Gly-Ser-Glu- ^■Thr-Thr

7 172 178 1 . 102 Gly-Lys-Arg-Gly-Gln-Asp-Ala

8 7 13 1 . 098 Val-Asp-Ser-Ser-Lys-Ser-Phe

9 232 238 1 . 097 Ser-Glu-Ser-Pro-Asn-Lys-Thr 10 68 74 1 . 086 Val-Asp-Asn-Glu-Asn-Pro-Leu

DAB_38g-(l-10)-IL2 was constructed from DAB_38g-IL-2. DAB_38g-IL-2 was constructed by removing a 309 bp Hpall - SphI restriction fragment from pDW24 and replacing it with oligonucleotide linker 261/274 (Table 4) to generate plasmid pDW27 (Fig. 1).

TABLE 4: Oligonucleotide Linkers.

construct oligonucleotide linker

DAB₃₈₉-IL-2 5'-CG-GGT-CAC-AAA-ACG-CAT-5' 1/2 HpaII-l/2SphI

CCA-GTG-TTT-TGC

This linker restores fragment B sequences from Pro383 to Thr387, and allows for in-frame fusion to IL-2 sequences at this position. Thus, in DAB_38g-IL-2 the 97 amino acids between Thr387 and His485 have been deleted. DAB_38g-(l-10)-IL-2 was constructed by inserting DNA encoding the spacer (see Table 1) into pDW27 at the SphI site, as described above. DAB_38g-(l-10)-IL-2 may also be generated directly from pDW24 (DAB₄₈₆-IL-2) by removal of the 309 bp Hpall-SphI fragment and replacing it with a linker that restores fragment B sequences from Pro383 to Thr387, encodes the polypeptide linker, and allows for the in-frame fusion to the IL-2 sequences.

The sequence of DT is given in Greenfield et al. (1983) Proc. Natl. Acad. Sci. USA 80_:6853-6857. The sequence encoding IL-2 was synthesized on an Applied

Biosystems DNA-Synthesizer, as described in Williams et al. (1988) Nucleic Acids Res. 1 :10453-10467, hereby incorporated by reference. The sequence of IL-2 is f-ound in Williams et al. (1988) Nucleic Acids Res. l6:10453-10467. Fusion of the sequence encoding mature DT to ATG using an oligonucleotide linker is described in Bishai et al. (1987) J. Bact. 169:5140-5151, hereby incorporated by reference. Chimeric toxins in which DT fragments were fused to murine interleukin 4 (IL-4), with or without the presence of a spacer were also constructed. DAB₄₈₆-IL-4, DAB₃₈₉-IL-4, and DAB_38g-(l-10) -IL-4 were constructed by methods known to those skilled in the art. Briefly,

DAB_38g-(l-10) -IL4 was constructed by first digesting a plasmid containing the

2

DAB_38g-(l-10) -IL-2 fusion (derived from pDW27) with

SphI and HindiII to remove the IL-2 coding sequence and then inserting a segment of DNA that encodes IL-4. The IL-4 encoding fragment includes linkers to allow insertion and in-frame fusion to the 3' end of the spacer encoding DNA. DAB_48g-IL-4 and DAB_38g-IL-4 were made by analogous treatment of pDW24 and pDW27, respectively. The sequence of murine IL-4 may be found in Lee et al. (1986) Proc. Natl. Acad. Sci. USA 83:2061-2065, hereby incorporated by reference. The sequence used in these constructions was obtained from DNAX (California). Oligonucleotides and nucleic acids were synthesized and manipulated as follows. Oligonucleotides were synthesized using cyanoethyl phosphoramidite chemistry on an Applied Biosystems 380A DNA synthesizer (Applied Biosystems Inc., Foster City, CA) . Following synthesis, oligonucleotides were purified by chromatography on Oligonucleotide Purification Cartridges (Applied Biosystems Inc., Foster City, CA) as directed by the manufacturer. Purified oligonucleotides were resuspended in TE buffer (10 mM Tris base, 1 mM EDTA, pH 8.0). To anneal complementary strands, equimolar concentrations of each strand were mixed in the presence of 100 mM NaCl, heated to 90°C for 10 min, and allowed to cool slowly to room temperature. Plasmid DNA was purified by the alkaline lysis/cesium chloride gradient method of Ausubel et al. (1989) Current Protocols in Molecular Biology, John Wiley & Sons, N.Y. DNA was digested with restriction endonucleases as recommended by the manufacturer (New England Biolabs, Beverly, MA and Bethesda Research

Laboratories, Gaithersburg, MD) . Restriction fragments for plasmid construction were extracted from agarose-TBE gels, ligated together (with or without oligonucleotide linkers) and used to transform E^ coli using standard methods. See Ausubel et al (1989), supra, and Maniatis et al. (1982), Molecular Cloning Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Plasmid DNA sequencing was performed according to the dideoxy chain termination method of Sanger et al. (1987) Proc. Natl. Acad. Sci. USA 74:5463-5467, as modified by Kraft et al. (1988) Bio Techniques 6 :544-547, using Sequenase (United States Biochemicals, Cleveland, OH). Expression and Purification of Chimeric Toxins

Expression and purification of chimeric toxins was as follows. All DT-related IL-2 fusion proteins used herein were expressed in the cytoplasm of E^ coli strain JM101 from the trc promoter, Amann et al. (1985), Gene 40_:183-190, hereby incorporated by reference. Recombinant E^ coli were grown in M9 minimal medium (Maniatis et al. (1982) supra) supplemented with

10 mg/ml casamino acids (Difco, Detroit, MI), 50 μg/ml ampicillin, and 0.5 ng/ml thymine in 10 liter volumes in a Microgen Fermentor (New Brunswick Scienctific, Edison, N.J.). Bacterial cultures were grown at 30°C, and sparged with air at 5 L/min. When the absorbance (A_5gonm) of the culture reached 0.3, expression of chimeric tox gene was induced by the addition of isopropyl-β-D- thiogalactopyranoside. Two hours after induction, bacteria were harvested by centrifugation, resuspended in buffer #101 (50 mM KH₂P0₄, 10 mM EDTA, 750 mM NaCl, 0.1% Tween 20, pH 8.0), and lysed by sonication (Branson Sonifier). Whole cells and debris were removed by centrifugation at 27,000 x g, and the clarified extract was then filter sterilized and applied to an anti-diphtheria toxin i munoaffinity column. Bound proteins were eluted with 4M guanidine hydrochloride, reduced by the addition of β-mercaptoethanol to 1% and then sized by high pressure liquid chromatography on a 7.5 x 600 mm G4000PW column (TosoHass). Prior to use, fusion toxins were exhaustively dialysed against HEPES buffered Hank's balanced salt solution (Gibco), pH 7.4. Purified diphtheria toxin was purchased from List Biological Laboratories (Campbell, CA) . The concentration of all purified proteins was determined by using Pierce Protein Assay reagent (Pierce Chemical Co., Rockford, IL) . Cytotoxicity

The dose response capacity of various chimeric toxins to block [ 14C]-leucine incorporation by HUT 102/6TG cells (which bear the high affinity IL-2 receptor) and YT2C2 cells (which bear only the intermediate affinity (p75) receptor) was determined.

Table 4 shows the concentration in moles of toxin reqnuiiirreedd to inhibit C 14C.]-leucine incorporation by 50% (IC₅₀). - 26

TABLE 5 CYTOTOXICITY

IC₅₀ (M)

TOXIN HUT102 YT2C2

IL-2

DAB₄₈₆-IL-2 120 X 10-12 70 X 10^-9

DAB3₈g-IL-2 40 x lO"¹² 11 X 10~⁹

DAB₃₈₉-(l-10)-IL-2 8 x 10~¹² 2 X 10~⁹

DAB₃₈₉-(l-10)²-IL-2 8 x 10-12 2 X 10~⁹

DAB₃₈₉-(l-10)³-IL-2 11 x lO----² 2.2 X 10^~9

As se~v~^ e 5, the toxicity of the DAB_38g - IL-2 chr B increased approximately 5-fold by the ad« acer peptide. The effect of two copies acer (DAB_38g-(l-10)²-IL-2) or

the (1-10) spacer (DAB_38g-(l-10)^,3-IL-2) have essentially the same effect as does one copy of the spacer (DAB_3gg-(l-10)-IL-2) .

Chimeric toxins in which DAB_4gg or DAB_3gg was fused to IL-4 (with or without the same spacer used on the IL-2 constructions, see Table 1) were also analyzed. When tested for cytotoxicity to

2

IL-4-receptor-bearing cells DAB_3gg-(l-10) -IL-4 was seen to be 2-10 times more cytotoxic than DAB_ggg-IL-4 which was seen to be about 10 times more cytotoxic than

^DAB486-^IL~⁴-

The addition of a spacer to fusions of DT fragments to melanocyte stimulating hormone had no effect on cytotoxicity.

Cytotoxicity assays were performed as follows.

For IL-2 chimeric toxins cultured HUT 102/6TG (Tsudo et al. (1986) Proc. Natl. Acad. Sci. USA 83:9694) or YT2C2 (Teshigawari et al. (1987) J. Exp. Med 165:223) cells were maintained in RPMI 1640 medium (Gibco, Grand Island, N.Y. ) supplemented with 10% fetal bovine serum (Cellect, GIBCO), 2 mM glutamine, and penicillin and streptomycin to 50 IU and 50 μg/ml, respectively. Cells were seeded in 96-well V-bottomed plates

(Linbro-Flow Laboratories, McLean, VA) at a concentration of 5 x 10 per well in complete medium. Toxins, or toxin-related materials, were added to varying concentrations (10 —12M to 10—6M) and the cultures were incubated for 18 hrs at 37°C in a 5% C0₂ atmosphere. Following incubation, the plates were centrifuged for 5 min. at 170 x g and the medium removed and replaced with 200 μl leucine-free medium (MEM, Gibco) containing 1.0 μCi/ml [ 14C]-leucine (New

England Nuclear, Boston, MA). After an additional 90 min. at 37°C, the plates were centrifuged for 5 min. at

170 x g, the medium was removed and the cells were lysed by the addition of 4 M KOH. Protein was precipitated by the addition of 10% trichloroacetic acid and the insoluble material was then collected on glass fiber filters using a cell harvester (Skatron, Sterling, VA) . Filters were washed, dried, and counted according to standard methods. Cells cultured with medium alone served as the control. IL-4 chimeric toxins were tested in a similar manner except that CT4R cells (William E.

Paul, NIH), P815 cells (ATCC) , or CTLL2 (ATCC) were

₄ seeded at 1 x 10 cells per well and incubated for 40 hours.

Competitive displacement experiments

Table 6 shows the competitive displacement of

[¹²⁵I.-labeled IL-2 from the high affinity IL-2 receptor (HUT102 cells) and the intermediate affinity

IL-2 receptor (YT2C2 cells) by various chimeric toxins. TABLE 6 Competitive Displacement

Toxin 50% displacement or HUT102 YT2C2 Control

-9

^DAB486-^IL-² 35 x 10-10 120 X 10 -9 ^DAB389-^IL-² 29 x 10-10 110 x 10

DAB₃₈₉-(l-10)-IL-2 13 x 10-10 30 X 10-9

DAB₃₈₉-(l-10)²-IL-2 12 x 10-10 28 x 10-9

DAB₃₈₉-(l-10)³-IL-2 13 x 10-10 32 x 10-9

As seen in Table 6, the binding affinity is increased by the insertion of the (1-10) spacer peptide into

^DAB389 -IL-2. Insertion of two copies of the spacer

(DAB_3gg-(l-10)'⁶-IL-2) or 3 copies of the spacer

(DAB_38g-(l-10) -IL-2) has essentially the same effect as does insertion of a single copy of the spacer

(DAB-,_fiQ-(1-10)-IL-2). Competitive displacement of [ 125I]-rIL-2 by recombinant IL-2 (rIL-2) and chimeric toxins was determined as follows. The radiolabeled IL-2 binding assay was performed essentially as described by Wang et al. (1987 J. Exp. Med. 166:1055-1069. Cells were harvested, washed with cell culture medium, and resuspended to 5 x 10 per ml and incubated with [ 125I]-rIL-2 (0.7 μCi/pmol) in the presence or absence of increasing concentrations of unlabeled rIL-2 or chimeric toxin for 30 min. at 37°C under 5% C0₂.

The reaction was then overlayed on a mixture of 80% 550 fluid (Accumetric Inc., Elizabethtown, KN) : 20% parafin oil (d = 1.03 g/ml) and microcentrifuged. The aqueous phase and the pellet of each sample, representing free and bound ligand, respectively, was then counted in a Nuclear Chicago gamma counter. Apparent dissociation constants, K,, were determined from the concentrations of unlabeled ligand required to displace 50% of radiolabeled rIL-2 binding to receptors. USE The improved chimeric toxins of the invention are administered to a mammal, e.g., a human, suffering from a medical disorder, e.g., cancer, or other conditions characterized by the presence of a class of unwanted cells to which a polypeptide ligand can selectively bind. The amount of protein administered will vary with the type of disease, extensiveness of the disease, and size of species of the mammal suffering from the disease. Generally, amounts will be in the range of those used for other cytotoxic agents used in the treatment of cancer, although in certain instances lower amounts will be needed because of the specificity and increased toxicity of the improved chimeric toxins.

The improved chimeric toxins can be admnistered using any conventional method; e.g., via injection, or via a timed-release implant. The improved chimeric toxins can be combined with any non-toxic, pharmaceutically-acceptable carrier substance.

Other embodiments are within the following claims. What is claimed is:

Claims

CLAIMS 1. A chimeric toxin including protein fragments joined together by peptide bonds, said chimeric toxin comprising, in sequential order, beginning at the amino terminal end of the chimeric toxin, (a) the enzymatically active Fragment A of diphtheria toxin, (b) a first fragment including the cleavage domain 1., adjacent said Fragment A of diphtheria toxin, (c) a second fragment comprising at least a portion of the hydrophobic transmembrane region of Fragment B of diphtheria toxin, said second fragment having a deletion of at least 50 diphtheria toxin amino acid residues, said deletion being C-terminal to said portion of the transmembrane region, and said second fragment not including domain 1^, (d) a spacer, (e) a portion of a cell-specific polypeptide ligand, said cell-specific polypeptide ligand being a cell growth factor, said portion including at least a portion of the binding domain of said polypeptide ligand, said portion of said binding domain being effective to cause said chimeric toxin to bind selectively to said target cell.

2. The chimeric toxin of claim 1, wherein said cell growth factor is a lymphokine.

3. The chimeric toxin of claim 1, wherein said deletion is at least 80 diphtheria toxin amino acid residues in length.

4. The chimeric toxin of claim 1, wherein said Fragment 3 of diphtheria toxin does not include any diphtheria toxin sequences C-terminal to amino acid residue 336 of native diphtheria toxin.

5. The chimeric toxin of claim l, wherein (a), (b), and (c) comprise DAB_38g.

6. The chimeric toxin of claim 1, wherein said spacer is at least 5 amino acids long.

7. The chimeric toxin of claim 1 wherein said spacer is 10-30 amino acids in length.

8. The chimeric toxin of claim 1, wherein said spacer, when placed between the sequence of DT fragment DAB₄₈₅ and amino acid residues 2-133 of IL-2, has a Bnorm value of 1.000 or g ^~reater.

9. The chimeric toxin of claim l, wherein said spacer, when placed between the sequence of DT fragment DAB._oc. and amino acid residues 2-133 of IL-2, has a Bnorm value of 1.125 or greater.

10. The chimeric toxin of claim 1, wherein said spacer, when placed between the sequence of DT fragment DAB₄₈₅ and amino acid residues 2-133 of IL-2, has a Bno„r_m„ value of 1.135 or greater.

11. The chimeric toxin of claim 1, wherein at least 60% of the amino acids in said spacer are from the group lysine, serine, glycine, proline, aspartic acid, glutamic acid, glutamine, threonine, asparagine, or axginine.

12. The chimeric toxin of claim 1, wherein at least 80% of the amino acids in said spacer are from the group lysine, serine, glycine, proline, aspartic acid, glutamic acid, glutamine, threonine, asparagine, or arginine.

13. The chimeric toxin of claim l, wherein said spacer is Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

14. The chimeric toxin of claim 1, wherein said spacer is at least 60% homologous to the sequence Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

15. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to the sequence Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

16. The chimeric toxin of claim 1, wherein said spacer is a multiple of Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

17. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to a multiple of Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

18. The chimeric toxin of claim 1, vherein said spacer is at least 60% homologous to a multiple of Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr.

19. The chimeric toxin of claim 1, wherein said spacer is Pro-Lys-Ser-Gly-Thr-Gln-Gly.

20. The chimeric toxin of claim 1, wherein said spacer is at least 60% homologous to the sequence Pro-Lys-Ser-Gly-Thr-Gln-Gly.

21. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to the sequence Pro-Lys-Ser-Gly-Thr-Gln-Gly.

22. The chimeric toxin of claim 1, wherein said spacer is a multiple of Pro-Lys-Ser-Gly-Thr-Gln-Gly.

23. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to a multiple of Pro-Lys-Ser-Gly-Thr-Gln-Gly.

24. The chimeric toxin of claim 1, wherein said spacer is at least 60% homologous to a multiple of Pro-Lys-Ser-Gly-Thr-Gln-Gly.

25. The chimeric toxin of claim 1, wherein said spacer is Pro-Thr-Ser-Ser-Ser-Thr-Lys.

26. The chimeric toxin of claim 1, wherein said spacer is at least 60% homologous to the sequence Pro-Thr-Ser-Ser-Ser-Thr-Lys.

27. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to the sequence Pro-Thr-Ser-Ser-Ser-Thr-Lys.

28. The chimeric toxin of claim 1, wherein said spacer is a multiple of Pro-Thr-Ser-Ser-Ser-Thr-Lys.

29. The chimeric toxin of claim 1, wherein said spacer is at least 80% homologous to a multiple of Pro-Thr-Ser-Ser-Ser-Thr-Lys.

30. The chimeric toxin of claim 1, wherein said spacer is at least 60% homologous to a multiple of Pro-Thr-Ser-Ser-Ser-Thr-Lys.

31. The chimeric toxin of claim 1, wherein said spacer results in an affinity of said chimeric toxin for said target cells that is greater than the affinity of a second chimeric toxin for said target cells, said second chimeric toxin being identical to said chimeric toxin except that said second chimeric toxin lacks said spacer.

32. The chimeric toxin of claim 1 wherein said spacer results in said chimeric toxin exhibiting cytotoxicity for said target cells that is a least 2 times greater than the cytotoxicity exhibited by a second chimeric toxin for said target cells, said second chimeric toxin being identical to said chimeric toxin except that said second chimeric toxin lacks said spacer.

33. The chimeric toxin of claim 1, wherein said portion of said polypeptide ligand is a portion of interleukin-2 effective to cause said chimeric toxin to bind to T cells.

34. The chimeric toxin of claim 1, wherein said portion of said polypeptide ligand is a portion of interleukin-4 effective to cause said chimeric toxin to bind to B cells.

35. The chimeric toxin of claim 1, wherein said chimeric toxin is encoded by a fused gene comprising regions coding for said protein fragments.

36. A DNA sequence encoding the chimeric toxin of claim 1.

37. An expression vector containing the DNA sequence of claim 35.

38. A cell transformed with the vector of claim 36.

39. A method of producing the chimeric toxin of claim 1 comprising culturing the cell of claim 37, and isolating said chimeric toxin from the cultured cell or supernatant.

40. A spacer peptide said spacer peptide having the sequence Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr, or tandem repeats thereof.

41. A DNA segment encoding the spacer peptide of claim 40.

42. The DNA segment of claim 41, further characterized in that it includes a linker at each end.

43. The peptide encoded by the DNA segment of claim 42.

44. A vector containing the DNA segment of claim 41.

45. A cell transformed with the vector of claim 44.

46. A method of making the spacer peptide of claim 40 comprising culturing the cell of claim 45, and isolating said spacer from the cultured cell or supernatant.

47. The DNA segment of claim 41, wherein said spacer is a tandem repeat of the subunit Ala-Pro-Thr-Ser-Ser-Ser-Thr-Lys-Lys-Thr, and the sequence encoding each occurence of said subunit in said spacer is nonhomologous with the sequence encoding every other occurence of said subunit in the molecule, said nonhomology being sufficient to prevent recombination between sequences encoding tandemly repeated subunits.

48. A method of preventing recombination between the tandemly repeated subunits of spacer-peptide-encoding DNA comprising choosing the codons of each subunit-encoding sequence such that the DNA encoding each subunit is nonhomologous with the DNA encoding every other subunit in the molecule, said nonhomology being sufficient to prevent recombination between tandemly repeated subunits.