US 5,677,274 AGrant
Anthrax Toxin Fusion Proteins and Related Methods
Issue Date:1997-10-14
•12 Claims
•1 Drawing Sheets
Abstract
The present invention provides a nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the anthrax protective antigen (PA) binding domain of the native anthrax lethal factor (LF) protein and a nucleotide sequence encoding an activity inducing domain of a second protein. Also provided is a nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the translocation domain and LF binding domain of the native anthrax PA protein and a nucleotide sequence encoding a ligand domain which specifically binds a cellular target. Proteins encoded by the nucleic acid of the invention, vectors comprising the nucleic acids and hosts capable of expressing the protein encoded by the nucleic acids are also provided. A composition comprising the PA binding domain of the native LF protein chemically attached to a non-LF activity inducing moiety is further provided. A method for delivering an activity to a cell is provided. The steps of the method include a) administering to the cell a protein comprising the translocation domain and the LF binding domain of the native PA protein and a ligand domain, and b) administering to the cell a product comprising the PA binding domain of the native LF protein and a non-LF activity inducing moiety, whereby the product administered in step b) is internalized into the cell and performs the activity within the cell. The invention also provides proteins including an anthrax protective antigen which has been mutated to replace the trypsin cleavage site with residues recognized specifically by the HIV-1 protease.
Metadata
Assignee
- The Government of the United States as represented by the Secretary of the Department of Health and Human Services
Inventors
- Stephen H. Leppla
- Kurt R. Klimpel
- Naveen Arora
- Yogendra Singh
- Peter J. Nichols
Application Information
Application Number:US 0828491
Filing Date:1993-06-25
Priority Date:1993-02-12
Art Unit:181
Classifications
IPC:
A61K 3900
Field of Search:
51442421.69
Patent Drawings (1 sheets)
Description
BACKGROUND OF THE INVENTION
The targeting of cytotoxic or other moieties to specific cell types has been proposed as a method of treating diseases such as cancer. Various toxins including Diphtheria toxin and Pseudomonas exotoxin A have been suggested as potential candidate toxins for this type of treatment. A difficulty of such methods has been the inability to selectively target specific cell types for the delivery of toxins or other active moieties.
One method of targeting specific cells has been to make fusion proteins of a toxin and a single chain antibody. A single-chain antibody (sFv) consists of an antibody light chain variable domain (V.sub.L) and heavy chain variable domain (V.sub.H), connected by a short peptide linker which allows the structure to assume a conformation capable of binding to antigen. In a diagnostic or therapeutic setting, the use of an sFv may offer attractive advantages over the use of a monoclonal antibody (MoAb). Such advantages include more rapid tumor penetration with concomitantly low retention in non-targeted organs (Yokota et al. Cancer Res 52:3402,1992), extremely rapid plasma and whole body clearance (resulting in high tumor to normal tissue partitioning) in the course of imaging studies (Colcher et al. Natl. Cancer Inst. 82:1191, 1990; Milenic et al. Cancer Res. 51:6363, 1991), and relatively low cost of production and ease of manipulation at the genetic level (Huston et al. Methods Enzymol. 203:46, 1991; Johnson, S. and Bird, R. E. Methods Enzymol. 203:88, 1991). In addition, sFv-toxin fusion proteins have been shown to exhibit enhanced anti-tumor activity in comparison with conventional chemically cross-linked conjugates (Chaudhary et al. Nature 339:394, 1989; Batra et al. Cell. Biol. 11:2200-2295, 1991). Among the first sFv to be generated were molecules capable of binding haptens (Bird et al. Science 242:423, 1988; Huston et al. Proc. Natl. Acad. Sci. USA 85:5879, 1988), cell-surface receptors (Chaudhary et al., 1989), and tumor antigens (Chaudhary et al. Proc. Natl. Acad. Sci. USA 87:1066, 1990; Colcher et al., 1990).
The gene encoding an sFv can be assembled in one of two ways: (i) by de novo construction from chemically synthesized overlapping oligonucleotides, or (ii) by polymerase chain reaction (PCR)-based cloning of V.sub.L and V.sub.H genes from hybridoma cDNA. The main disadvantages of the first approach are the considerable expense involved in oligonucleotide synthesis, and the fact that the sequence of V.sub.L and V.sub.H must be known before gene assembly is possible. Consequently, the majority of the sFv reported to date were generated by cloning from hybridoma cDNA; nevertheless, this approach also has inherent disadvantages, because it requires availability of the parent hybridoma or myeloma cell line, and problems are often encountered when attempting to retrieve the correct V region genes from heterologous cDNA. For example, hybridomas in which the immortalizing fusion partner is derived from MOPC-21 may express a V.sub.L kappa transcript which is aberrantly rearranged at the VJ recombination site, and which therefore encodes a non-functional light chain (Cabilly & Riggs, 1985; Carroll et al., 1988). Cellular levels of this transcript may exceed that generated from the productive V.sub.L gene, so that a large proportion of the product on PCR amplification of hybridoma cDNA will not encode a functional light chain. A second disadvantage of the PCR-based method, frequently encountered by the inventors, is the variable success of recovering V.sub.H genes using the conditions so far reported in the literature, presumably because the number of mismatches between primers and the target sequence destabilizes the hybrid to an extent which inhibits PCR amplification.
Thus, methods of targeting toxins to specific cells using single-chain antibodies methods have been difficult to practice because of the difficulties in obtaining single chain antibodies and other targeting moieties. Also, none of the proposed treatment methods has been fully successful, because of the need to fuse the toxin to the targeting moiety, thus disrupting either the toxin function or the targeting function. Thus, a need exists for a means to target molecules having a desired activity to a specific cell population.
Bacterial and plant protein toxins have evolved novel and efficient strategies for penetrating to the cytosol of mammalian cells, and this ability has been exploited to develop anti-tumor and anti-HIV cytotoxic agents. Examples include ricin and Pseudomonas exotoxin A (PE) chimeric toxins and immunotoxins.
Pseudomonas exotoxin A (PE) is a toxin for which a detailed analysis of functional domains exists. The sequence is deposited with GenBank. Structural determination by X-ray diffraction, expression of deleted proteins, and extensive mutagenesis studies have defined three functional domains in PE: a receptor-binding domain (residues 1-252 and 365-399) designated Ia and Ib, a central translocation domain (amino acids 253-364, domain II), and a carboxyl-terminal enzymatic domain (amino acids 400-613, domain III). Domain III catalyzes the ADP-ribosylation of elongation factor 2 (EF-2), which results in inhibition of protein synthesis and cell death. Recently it was also found that an extreme carboxyl terminal sequence is essential for toxicity (Chaudhary et al. Proc. Natl. Acad. Sci. U.S.A. 87:308-312, 1990; Seetharam et al. J. Biol. Chem. 266:17376-17381, 1991). Since this sequence is similar to the sequence that specifies retention of proteins in the endoplasmic reticulum (ER) (Munro, S. and Pelham, H. R. B. Cell 48:899-907, 1987), it was suggested that PE must pass through the ER to gain access to the cytosol. Detailed knowledge of the structure of PE has facilitated use of domains II, Ib, and III (together designated PE40) in hybrid toxins and immunotoxins.
Bacillus anthracis produces three proteins which when combined appropriately form two potent toxins, collectively designated anthrax toxin. Protective antigen (PA, 82,684 Da (Dalton) (SEQ ID NOS: 3 and 4)) and edema factor (EF, 89,840 Da) combine to form edema toxin (ET), while PA and lethal factor (LF, 90,237 Da (SEQ ID NOS: 1 and 2)) combine to form lethal toxin (LT) (Leppla, S. H. Alouf, J. E. and Freer, J. H., eds. Academic Press, London 277-302, 1991). ET and LT each conform to the AB toxin model, with PA providing the target cell binding (B) function and EF or LF acting as the effector or catalytic (A) moieties. A unique feature of these toxins is that LF and EF have no toxicity in the absence of PA, apparently because they cannot gain access to the cytosol of eukaryotic cells.
The genes for each of the three anthrax toxin components have been cloned and sequenced (Leppla, 1991). This showed that LF and EF have extensive homology in amino acid residues 1-300. Since LF and EF compete for binding to PA63, it is highly likely that these amino-terminal regions are responsible for binding to PA63. Direct evidence for this was provided in a recent mutagenesis study (Quinn et al. J. Biol. Chem. 266:20124-20130, 1991); all mutations made within amino acid residues 1-210 of LF led to decreased binding to PA63. The same study also suggested that the putative catalytic domain of LF included residues 491-776 (Quinn et al., 1991). In contrast, the location of functional domains within the PA63 polypeptide is not obvious from inspection of the deduced amino acid sequence. However, studies with monoclonal antibodies and protease fragments (Leppla, 1991) and subsequent mutagenesis studies (Singh et al. J. Biol. Chem. 266:15493-15497, 1991) showed that residues at and near the carboxyl terminus of PA are involved in binding to receptor.
PA is capable of binding to the surface of many types of cells. After PA binds to a specific receptor (Leppla, 1991) on the surface of susceptible cells, it is cleaved at a single site by a cell surface protease, probably furin, to produce an amino-terminal 19-kDa fragment that is released from the receptor/PA complex (Singh et al. J. Biol. Chem. 264:19103-19107, 1989). Removal of this fragment from PA exposes a high-affinity binding site for LF and EF on the receptor-bound 63-kDa carboxyl-terminal fragment (PA63). The complex of PA63 and LF or EF enters cells and probably passes through acidified endosomes to reach the cytosol.
Cleavage of PA occurs after residues 164-167, Arg-Lys-Lys-Arg(SEQ. ID NO: 32). This site is also susceptible to cleavage by trypsin and can be referred to as the trypsin cleavage site. Only after cleavage is PA able to bind either EF or LF to form either ET or LT.
Prior work had shown that the carboxyl terminal PA fragment (PA63) can form ion conductive channels in artificial lipid membranes (Blaustein et al. Proc. Natl. Acad. Sci. U.S.A. 86:2209-2213, 1989; Koehler, T. M. and Collier, R. J. Mol. Microbiol. 5:1501-1506, 1991), and that LF bound to PA63 on cell surface receptors can be artificially translocated across the plasma membrane to the cytosol by acidification of the culture medium (Friedlander, A. M. J. Biol. Chem. 261:7123-7126, 1986). Furthermore, drugs that block endosome acidification protect cells from LF (Gordon et al. J. Biol. Chem. 264:14792-14796, 1989; Friedlander, 1986; Gordon et al. Infect. Immun. 56:1066-1069, 1988). The mechanisms by which EF is internalized have been studied in cultured cells by measuring the increases in cAMP concentrations induced by PA and EF (Leppla, S. H. Proc. Natl. Acad. Sci. U.S.A. 79:3162-3166, 1982; Gordon et al., 1989). However, because assays of cAMP are relatively expensive and not highly precise, this is not a convenient method of analysis. Internalization of LF has been analyzed only in mouse and rat macrophages, because these are the only cell types lysed by the lethal toxin.
SUMMARY OF THE INVENTION
The present invention provides a nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the PA binding domain of the native LF protein and a nucleotide sequence encoding an activity inducing domain of a second protein. Also provided is a nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the translocation domain and LF binding domain of the native PA protein and a nucleotide sequence encoding a ligand domain which specifically binds a cellular target. Proteins encoded by the nucleic acid of the invention, vectors comprising the nucleic acids and hosts capable of expressing the protein encoded by the nucleic acids are also provided.
A composition comprising the PA binding domain of the native LF protein chemically attached to an activity inducing moiety is further provided.
A method for delivering an activity to a cell is provided. The steps of the method include administering to the cell (a) a protein comprising the translocation domain and the LF binding domain of the native PA protein and a ligand domain and (b) a product comprising the PA binding domain of the native LF protein and a non-LF activity inducing moiety, whereby the product administered in step (b) is internalized into the cell and performs the activity within the cell.
Characteristics unique to anthrax toxin are exploited to make novel cell-specific cytotoxins. A site in the PA protein of the toxin which must be proteolytically cleaved for the activity-inducing moiety of the toxin to enter the cell is replaced by the consensus sequence recognized by a specific protease. Thus, the toxin will only act on cells infected with intracellular pathogens which make that specific protease.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a graph of the percent to which mutant proteins are cleaved by purified HIV-1 protease. The mutant proteins include protective antigen (PA) mutated to include the HIV-1 protease cleavage site in place of the natural trypsin cleavage site.
DESCRIPTION OF THE PREFERRED EMBODIMENT
Nucleic Acids
Lethal Factor (LF)
The present invention provides an isolated nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the PA binding domain of the native LF protein and a nucleotide sequence encoding an activity inducing domain of a second protein. The LF gene and native LF protein are shown in SEQ. ID NO: 1 and 2, respectively. The PA gene and native PA protein are shown in SEQ. ID NO: 3 and 4, respectively.
The second protein can be a toxin, for example Pseudomonas exotoxin A (PE), the A chain of Diphtheria toxin or shiga toxin. The activity inducing domains of numerous other known toxins can be included in the fusion protein encoded by the presently claimed nucleic acid. The activity inducing domain need not be a toxin, but can have other activities, including but not limited to stimulating or reducing growth, selectively inhibiting DNA replication, providing enzymatic activity or providing a source of radiation. In any case, the fusion proteins encoded by the nucleic acids of the present invention must be capable of being internalized and capable of expressing the specified activity in a cell. A given LF fusion protein of the present invention can be tested for its ability to be internalized and to express the desired activity using methods as described herein, particularly in Examples 1 and 2.
An example of a nucleic acid of the invention comprises the nucleotide sequence defined in the Sequence Listing as SEQ. ID NO: 5. This nucleic acid encodes a fusion of LF residues 1-254 with the two-residue linker "TR" and PE residues 401-602 (SEQ. ID NO: 6). The protein includes a Met-Val-Pro- sequence at the beginning of the LF sequence. Means for obtaining this fusion protein are further described below and in Example 1.
A further example of a nucleic acid of this invention comprises the nucleotide sequence defined in the Sequence Listing as SEQ. ID NO: 7. This nucleic acid encodes a fusion of LF residues 1-254 with the two-residue linker "TR" and PE residues 398-613. (SEQ. ID NO: 8) The junction point containing the "TR" is the sequence LTRA and the Met-Val-Pro- is also present. This fusion protein and methods for obtaining it are further described below and in Example 2.
Another example of the nucleic acid of the present invention comprises the nucleotide sequence defined in the Sequence Listing as SEQ. ID NO: 9. This nucleic acid encodes a fusion of LF residues 1-254 with the two residue linker and PE residues 362-613. (SEQ. ID NO: 10) This fusion protein is further described in Example 1.
Alternatively, the nucleic acid can include the entire coding sequence for the LF protein fused to a non-LF activity inducing domain. Other LF fusion proteins of various sizes and methods of making and testing them for the desired activity are also provided herein, particularly in Examples 1 and 2.
Protective Antigen (PA)
Also provided is an isolated nucleic acid encoding a fusion protein comprising a nucleotide sequence encoding the translocation domain and LF binding domain of the native PA protein and a nucleotide sequence encoding a ligand domain which specifically binds a cellular target.
An example of a nucleic acid of this invention comprises the nucleotide sequence defined in the Sequence Listing as SEQ. ID NO: 11. This nucleic acid encodes a fusion of PA residues 1-725 and human CD4 residues 1-178, the portion which binds to gp120 exposed on HIV-1 infected cells (SEQ ID NO:12). This fusion protein and methods for obtaining and testing fusion proteins are further described below and in Examples 3, 4 and 5.
The PA fusion protein encoding nucleic acid provided can encode any ligand domain that specifically binds a cellular target, e.g. a cell surface receptor, an antigen expressed on the cell surface, etc. For example, the nucleic acid can encode a ligand domain that specifically binds to an HIV protein expressed on the surface of an HIV-infected cell. Such a ligand domain can be a single chain antibody which is expressed as a fusion protein as provided above and in Examples 3, 4 and 5. Alternatively, the nucleic acid can encode, for example, a ligand domain that is a growth factor, as provided in Example 3.
Although the PA encoding sequence of the nucleic acid encoding the PA fusion proteins of this invention need only include the nucleotide sequence encoding the translocation domain and LF binding domain of the native PA protein, the nucleic acid can further comprise the nucleotide sequence encoding the remainder of the native PA protein. Any sequences to be included beyond those required, can be determined based on routine considerations such as ease of manipulation of the nucleic acid, ease of expression of the product in the host, and any effect on translocation/internalization as taught in the examples.
Proteins
Proteins encoded by the nucleic acids of the present invention are also provided.
LF Fusion Proteins
The present invention provides LF fusion proteins encoded by the nucleic acids of the invention as described above and in the examples. Specifically, fusions of the LF gene with domains II, Ib, and III of PE can be made by recombinant methods to produce in-frame translational fusions. Recombinant genes (e.g., SEQ ID NOs: 5, 7 and 9) were expressed in Escherichia coli (E. coli), and the purified proteins were tested for activity on cultured cells as provided in Examples 1 and 2. Certain fusion proteins are efficiently internalized via the PA receptor to the cytosol. These examples demonstrate that this system can be used to deliver many different polypeptides into targeted cells.
Although specific examples of these proteins are provided, given the present teachings regarding the preparation of LF fusion proteins, other embodiments having other activity inducing domains can be practiced using routine skill.
Using current methods of genetic manipulation, a variety of other activity inducing moieties (e.g., polypeptides) can be translated as fusion proteins with LF which in turn can be internalized by cells when administered with PA or PA fusion proteins. Fusion proteins generated by this method can be screened for the desired activity using the methods set forth in the Examples and by various routine procedures. Based on the data presented here, the present invention provides a highly effective system for delivery of an activity inducing moiety into cells.
PA fusion proteins
The present invention provides PA fusion proteins encoded by the nucleic acids of the invention. Specifically fusions of PA with single chain antibodies and CD4 are provided.
Using current methods of genetic manipulation, a variety of other ligand domains (e.g., polypeptides) can be translated as fusion proteins with PA which in turn can specifically target cells and facilitate internalization LF or LF fusion proteins. Based on the data presented here, the present invention provides a highly effective system for delivery of an activity inducing moiety into a particular type or class of cells.
Although specific examples of these proteins are provided, given the present teachings regarding the preparation of PA fusion proteins, other embodiments having other ligand domains can be practiced using routine skill. The fusion proteins generated can be screened for the desired specificity and activity utilizing the methods set forth in the example and by various routine procedures. In any case, the PA fusion proteins encoded by the nucleic acids of the present invention must be able to specifically bind the selected target cell, bind LF or LF fusions or conjugates and internalize the LF fusion/conjugate.
Conjugates
A composition comprising the PA binding domain of the native LF protein chemically attached to an activity inducing moiety is provided. Such an activity inducing moiety is an activity not present on native LF. The composition can comprise an activity inducing moiety that is, for example, a polypeptide, a radioisotope, an antisense nucleic acid or a nucleic acid encoding a desired gene product.
Using current methods of chemical manipulation, a variety of other moieties (e.g., polypeptides, nucleic acids, radioisotopes, etc.) can be chemically attached to LF and can be internalized into cells and can express their activity when administered with PA or PA fusion proteins. The compounds can be tested for the desired activity and internalization following the methods set forth in the Examples. For example, the present invention provides an LF protein fragment 1-254 (LF1-254) with a cysteine residue added at the end of LF1-254 (LF1-254Cys). Since there are no other cysteines in LF, this single cysteine provides a convenient attachment point through which to chemically conjugate other proteins or non-protein moieties.
Vectors and Hosts
A vector comprising the nucleic acids of the present invention is also provided. The vectors of the invention can be in a host capable of expressing the protein encoded by the nucleic acid.
To express the proteins and conjugates of the present invention, the nucleic acids can be operably linked to signals that direct gene expression. A nucleic acid is "operably linked" when it is placed into a functional relationship with another nucleic acid sequence. For instance, a promoter or enhancer is operably linked to a coding sequence if it affects the transcription of the sequence. Generally, operably linked means that the nucleic acid sequences being linked are contiguous and, where necessary to join two protein coding regions, contiguous and in reading frame.
The gene encoding a protein of the invention can be inserted into an "expression vector", "cloning vector", or "vector," terms which usually refer to plasmids or other nucleic acid molecules that are able to replicate in a chosen host cell. Expression vectors can replicate autonomously, or they can replicate by being inserted into the genome of the host cell. Vectors that replicate autonomously will have an origin of replication or autonomous replicating sequence (ARS) that is functional in the chosen host cell(s). Often, it is desirable for a vector to be usable in more than one host cell, e.g., in E. coli for cloning and construction, and in a mammalian cell for expression.
The particular vector used to transport the genetic information into the cell is also not particularly critical. Any of the conventional vectors used for expression of recombinant proteins in prokaryotic or eukaryotic cells can be used.
The expression vectors typically have a transcription unit or expression cassette that contains all the elements required for the expression of the DNA encoding a protein of the invention in the host cells. A typical expression cassette contains a promoter operably linked to the DNA sequence encoding the protein, and signals required for efficient polyadenylation of the transcript. The promoter is preferably positioned about the same distance from the heterologous transcription start site as it is from the transcription start site in its natural setting. As is known in the art, however, some variation in this distance can be accommodated without loss of promoter function.
The DNA sequence encoding the protein of the invention can be linked to a cleavable signal peptide sequence to promote secretion of the encoded protein by the transformed cell. Additional elements of the vector can include, for example, selectable markers and enhancers. Selectable markers, e.g., tetracycline resistance or hygromycin resistance, permit detection and/or selection of those cells transformed with the desired DNA sequences (see, e.g., U.S. Pat. No. 4,704,362).
Enhancer elements can stimulate transcription up to 1,000 fold from linked homologous or heterologous promoters. Many enhancer elements derived from viruses have a broad host range and are active in a variety of tissues. For example, the SV40 early gene enhancer is suitable for many cell types. Other enhancer/promoter combinations that are suitable for the present invention include those derived from polyoma virus, human or murine cytomegalovirus, the long terminal repeat from various retroviruses such as murine leukemia virus, murine or Rous sarcoma virus, and HIV. See, Enhancers and Eukaryotic Expression, Cold Spring Harbor Pres, Cold Spring Harbor, N.Y. 1983, which is incorporated herein by reference.
In addition to a promoter sequence, the expression cassette should also contain a transcription termination region downstream of the structural gene to provide for efficient termination. The termination region can be obtained from the same gene as the promoter sequence or can be obtained from a different gene.
For more efficient translation in mammalian cells of the mRNA encoded by the structural gene, polyadenylation sequences are also commonly added to the vector construct. Two distinct sequence elements are required for accurate and efficient polyadenylation: GU or U rich sequences located downstream from the polyadenylation site and a highly conserved sequence of six nucleotides, AAUAAA, located 11-30 nucleotides upstream. Termination and polyadenylation signals that are suitable for the present invention include those derived from SV40, or a partial genomic copy of a gene already resident on the expression vector.
The vectors containing the gene encoding the protein of the invention are transformed into host cells for expression. "Transformation" refers to the introduction of vectors containing the nucleic acids of interest directly into host cells by well known methods. The particular procedure used to introduce the genetic material into the host cell for expression of the protein is not particularly critical. Any of the well known procedures for introducing foreign nucleotide sequences into host cells can be used. It is only necessary that the particular procedure utilized be capable of successfully introducing at least one gene into the host cell which is capable of expressing the gene.
Transformation methods, which vary depending on the type of host cell, include electroporation; transfection employing calcium chloride, rubidium chloride calcium phosphate, DEAE-dextran, or other substances; microprojectile bombardment; lipofection; infection (where the vector is an infectious agent); and other methods. See, generally, Sambrook et al., (1989) supra, and Current Protocols in Molecular Biology, supra. Reference to cells into which the nucleic acids described above have been introduced is meant to also include the progeny of such cells.
There are numerous prokaryotic expression systems known to one of ordinary skill in the art useful for the expression of the antigen. E. coli is commonly used, and other microbial hosts suitable for use include bacilli, such as Bacillus subtilus, and other enterobacteriaceae, such as Salmonella, Serratia, and various Pseudomonas species. One can make expression vectors for use in these prokaryotic hosts; the vectors will typically contain expression control sequences compatible with the host cell (e.g., an origin of replication, a promoter). Any number of a variety of well-known promoters can be used, such as the lactose promoter system, a tryptophan (Trp) promoter system, a beta-lactamase promoter system, or a promoter from phage lambda. The promoters will typically control expression, optionally with an operator sequence, and have ribosome binding site sequences, for example, for initiating and completing transcription and translation. If necessary, an amino terminal methionine can be provided by insertion of a Met codon 5' and in-frame with the codons for the protein. Also, the carboxy-terminal end of the protein can be removed using standard oligonucleotide mutagenesis procedures, if desired.
Host bacterial cells can be chosen that are mutated to be reduced in or free of proteases, so that the proteins produced are not degraded. For Bacillus expression systems in which the proteins are secreted into the culture medium, strains are available that are deficient in secreted proteases.
Mammalian cell lines can also be used as host cells for the expression of polypeptides of the invention. Propagation of mammalian cells in culture is per se well known. See, Tissue Culture, Academic Press, Kruse and Patterson, ed. (1973). Host cell lines may also include such organisms as bacteria (e.g., E. coli or B. subtills), yeast, filamentous fungi, plant cells, or insect cells, among others.
Purification of Protein
After standard transfection or transformation methods are used to produce prokaryotic, mammalian, yeast, or insect cell lines that express large quantities of the protein of the invention, the protein is then purified using standard techniques which are known in the art. See, e.g., Colley et al. (1989) J. Biol. Chem. 64: 17619-17622; and Methods in Enzymology, "Guide to Protein Purification", M. Deutscher, ed. Vol. 182 (1990).
Standard procedures of the art that can be used to purify proteins of the invention include ammonium sulfate precipitation, affinity and fraction column chromatography, gel electrophoresis and the like. See, generally, Scopes, R., Protein Purification, Springer-Verlag, N.Y. (1982), and U.S. Pat. No. 4,512,922 disclosing general methods for purifying protein from recombinantly engineered bacteria.
If the expression system causes the protein of the invention to be secreted from the cells, the recombinant cells are grown and the protein is expressed, after which the culture medium is harvested for purification of the secreted protein. The medium is typically clarified by centrifugation or filtration to remove cells and cell debris and the proteins can be concentrated by adsorption to any suitable resin such as, for example, CDP-Sepharose, asialoprothrombin-Sepharose 4B, or Q Sepharose, or by use of ammonium sulfate fractionation, polyethylene glycol precipitation, or by ultrafiltration. Other means known in the art are equally suitable. Further purification of the protein can be accomplished by standard techniques, for example, affinity chromatography, ion exchange chromatography, sizing chromatography, or other protein purification techniques used to obtain homogeneity. The purified proteins are then used to produce pharmaceutical compositions, as described below.
Alternatively, vectors can be employed that express the protein intracellularly, rather than secreting the protein from the cells. In these cases, the cells are harvested, disrupted, and the protein is purified from the cellular extract, e.g., by standard methods. If the cell line has a cell wall, then initial extraction in a low salt buffer may allow the protein to pellet with the cell wall fraction. The protein can be eluted from the cell wall with high salt concentrations and dialyzed. If the cell line glycosolates the protein, then the purified glycoprotein may be enhanced by using a Con A column. Anion exchange columns Mono Q.RTM. Pharmacia) and gel filtration columns may be used to further purify the protein. A highly purified preparation can be achieved at the expense of activity by denaturing preparative polyacrylamide gel electrophoresis.
Protein analogs can be produced in multiple conformational forms which are detectable under nonreducing chromatographic conditions. Removal of those species having a low specific activity is desirable and is achieved by a variety of chromatographic techniques including anion exchange or size exclusion chromatography.
Recombinant analogs can be concentrated by pressure dialysis and buffer exchanged directly into volatile buffers (e.g., N-ethylmorpholine (NEM), ammonium bicarbonate, ammonium acetate, and pyridine acetate). In addition, samples can be directly freeze-dried from such volatile buffers resulting in a stable protein powder devoid of salt and detergents. In addition, freeze-dried samples of recombinant analogs can be efficiently resolubilized before use in buffers compatible with infusion (e.g., phosphate buffered saline). Other suitable buffers might include hydrochloride, hydrobromide, sulphate acetate, benzoate, malate, citrate, glycine, glutamate, and aspartate.
Specific Embodiments
Toxins Modified to Contain Intracellular Pathogen Protease Recognition Sites
One aspect of the invention exploits the fact that PA and other toxins must be proteolytically cleaved in order to acquire activity, in conjunction with the fact that some cells infected with an intracellular pathogen possess an active protease that has a relatively narrow substrate specificity (for example, HIV-infected cells). The protease site found in the native toxin is replaced with an intracellular pathogen specific protease site. Thus, the protease in cells that are infected by the intracellular pathogen cleaves the modified toxin, which then becomes active and kills the cell.
Intracellular pathogens that can be targeted by the products and methods of the present invention include any pathogen that produces a protease having a specific recognition site. Such pathogens can include prokaryotes (including rickettsia, Mycobacterium tuberculosis, etc.), mycoplasma, eukaryotic pathogens (e.g. pathogenic fungi, etc.), and viruses. One example of an intracellular pathogen that produces a specific protease is human immunodeficiency virus (HIV). The HIV-1 protease cleaves viral polyproteins to generate functional structural proteins as well as the reverse transcriptase and the protease itself. HIV-1 replication and viral infectivity are absolutely dependent on the action of the HIV-1 protease.
An intracellular pathogen specific protease site can be introduced into any natural or recombinant toxin for which proteolytic cleavage is required for toxicity. For example, one can replace the anthrax PA trypsin cleavage site (R164-167) of PA with the HIV-1 protease site. Alternatively, the diphtheria toxin disulfide loop sequence (see O'Hare, et al. FEBS 273 (1, 2): 200-204 (October 1990)) can be replaced with the HIV-1 protease cleavage site in order to obtain a toxin specific to HIV-1 infected cells. Similarly, the normally occurring diphtheria toxin sequence at residues 191-194 (Williams, et al. J. Biol. Chem. 265(33): 20673-20677 (1990)) can be replaced by an intracellular pathogen specific protease site such as the HIV-1 protease cleavage sequence. The DAB486-IL-2 fusion toxin of Williams and the improved DAB389-IL-2 toxin are effective on HIV-1 infected cells, which express high levels of the IL-2 receptor. Williams, J. Biol. Chem. 265:20673. Addition of the HIV-1 protease cleavage site would provide a further degree of specificity. Similarly, the botulinum toxin C2 toxin is like the anthrax toxin in requiring a cleavage within a native protein subunit (see Ohishi and Yanagimoto, Infection and Immunity 60(11): 4648-4655 (November 1992)), so it too can be made specific for cells infected by an intracellular pathogen such as HIV-1.
In one embodiment of the invention, the protease site of PA is replaced by the site recognized by the HIV-1 protease. The cellular protease that cleaves PA absolutely requires the presence of the Arg 164 and Arg 167 residues, because replacement of either residue yields a PA molecule which is not cleaved after binding to the cell surface. However, any PA substitution mutant which retains at least one Arg or Lys residue within residues 164-167 can be activated by treatment with trypsin. Because the PA63 fragments produced by trypsin digestion have a variety of different amino terminal residues, it is clear that there is not a strict constraint on the identity of the terminal residues. Klimpel, et al. , Proc. Natl. Acad. Sci. 89:10277-10281 (1992).
Replacement of residues 164-167 of PA with residues that match the HIV-1 protease recognition site can render exogenously added PA inactive on cells which do not possess the HIV-1 protease. However, those cells that do express the HIV-1 protease (i.e., cells infected with HIV-1 or cells engineered to produce the protease) would cleave and thereby activate the mutant PA. The activated PA proteins can then bind and internalize cytotoxic fusion proteins, such as LF-PE, added exogenously.
Based on extensive studies of the substrate specificity of the protease, several PA variants were designed and produced which relate to the invention. These are shown below, with the residues underlined between which the cleavage occurred. PA proteins which have been mutated to replace R164-167 with an amino acid sequence recognized by the HIV-1 protease are referred to as "PAHIV. " (SEQ. ID NO: 13)
The mutant proteins PAHIV#(1-4) were cleaved correctly by the HIV-1 protease.
Table 1 shows the amino acids and their corresponding abbreviations and symbols.
Preferably, the mutations at R164-167 of PA are accomplished by cassette mutagenesis, although other methods are feasible as discussed below. In summary, three pieces of DNA are joined together. The first piece has vector sequences and encodes the "front half" (5' end of the gene) of PA protein, the second is a short piece of DNA (a cassette) and encodes a small middle piece of PA protein and the third encodes the "back half" (3' end of the gene) of PA. The cassette contains codons for the amino acids that are required to complete the cleavage site for the intracellular pathogen protease. This method was used to make mutants in the plasmid pYS5 although other plasmids could be employed.
Alternatively, the mutations can be accomplished by use of the polymerase chain reaction (PCR) and other methods as discussed below. PCR duplicates a segment of DNA many times, resulting in an amplification of that segment. The reaction produces enough of the segment of DNA so that it can be modified with restriction enzymes and cloned. During the reaction a synthetic oligonucleotide primer is used to start the duplication of the target DNA segment. Each synthetic primer can be designed to introduce novel DNA sequences into the DNA molecule, or to change existing DNA sequences.
Modification of Toxins to Broaden or Alter Target Cell Specificity
Another aspect of the invention involves compounds and methods for broadening or changing the range of cell types against which a toxin is effective. For example, the lethal anthrax toxin, PA+LF, is acutely toxic to mouse macrophage cells, apparently due to the specific expression in these cells of a target for the catalytic activity of LF. Other cell types are not affected by LF. However, in the present invention, LF is used to construct cytotoxins having broad cell specificity.
A detailed analysis of the domains of LF identified the amino-terminal 254 amino acids as the region that binds to PA63. Fusion proteins containing residues 1-254 of LF and the ADP-ribosylation domain of Pseudomonas exotoxin A (PE) were designed according to the invention. These fusion proteins are highly toxic to cultured cells, but only when PA is administered simultaneously.
Synthesis of Genes that Encode Proteins of the Invention
Genes that encode toxins having altered protease recognition sites or fusion proteins having a binding domain from one protein and an activity inducing domain of a second protein can be synthesized by methods known to those skilled in the art. As an example of techniques that can be utilized, the synthesis of genes encoding modified anthrax toxin subunits LF and PA are now described.
The DNA sequences for native PA and LF are known. Knowledge of these DNA sequences facilitates the preparation of genes and can be used as a starting point to construct DNA molecules that encode mutants of PA and/or LF. The protein mutants of the invention are soluble and include internal amino acid substitutions. Furthermore, these mutants are purified from, or secreted from, cells that have been transfected or transformed with plasmids containing genes which encode these proteins. Methods for making modifications, such as amino acid substitutions, deletions, or the addition of signal sequences to cloned genes are known. Specific methods used herein are described below.
The gene for PA or LF can be prepared by several methods. Genomic and cDNA libraries are commercially available. Oligonucleotide probes, specific to the desired gene, can be synthesized using the known gene sequence. Methods for screening genomic and cDNA libraries with oligonucleotide probes are known. A genomic or cDNA clone can provide the necessary starting material to construct an expression plasmid for the desired protein using known methods.
A protein encoding DNA fragment can be cloned by taking advantage of restriction endonuclease sites which have been identified in regions which flank or are internal to the gene. See Sambrook, et al., Molecular Cloning: A Laboratory Manual 2d.ed. Cold Spring Harbor Laboratory Press (1989), "Sambrook" hereinafter.
Genes encoding the desired protein can be made from wild-type genes constructed using the gene encoding the full length protein. One method for producing wild-type genes for subsequent mutation combines the use of synthetic oligonucleotide primers with polymerase extension on a mRNA or DNA template. This PCR method amplifies the desired nucleotide sequence. U.S. Pat. Nos. 4,683,195 and 4,683,202 describe this method. Restriction endonuclease sites can be incorporated into the primers. Genes amplified by PCR can be purified from agarose gels and cloned into an appropriate vector. Alterations in the natural gene sequence can be introduced by techniques such as in vitro mutagenesis and PCR using primers that have been designed to incorporate appropriate mutations.
The proteins described herein can be expressed intracellularly and purified, or can be secreted when expressed in cell culture. If desired, secretion can be obtained by the use of the native signal sequence of the gene. Alternatively, genes encoding the proteins of the invention can be ligated in proper reading frame to a signal sequence other than that corresponding to the native gene. Though the PA recombinant proteins of the invention are typically expressed in B. anthracis, they can be expressed in other hosts, such as E. coli.
The proteins of this invention are described by their amino acid sequences and by their nucleotide sequence, it being understood that the proteins include their biological equivalents such that this invention includes minor or inadvertent substitutions and deletions of amino acids that have substantially little impact on the biological properties of the analogs. In some circumstances it may be feasible to substitute rare or non-naturally occurring amino acids for one or more of the twenty common amino acids listed in Table 2. Examples include ornithine and acetylated or hydroxylated forms. See generally Stryer, L., Biochemistry 3d ed. (1988).
Alternative nucleotide sequences can be used to express analogs in various host cells. Furthermore, due to the degeneracy of the genetic code, equivalent codons can be substituted to encode the same polypeptide sequence. Additionally, sequences (nucleotide and amino acid) with substantial identity to those of the invention are also included. Identity in this sense means the same identity (of base pair or amino acid) and order (of base pairs or amino acids). Substantial identity includes entities that are greater than 80% identical. Preferably, substantial identity refers to greater than 90% identity. More preferably, it refers to greater than 95% identity.
Mutagenesis
Mutagenesis can be performed to yield point mutations, deletions, or insertions to alter the specific regions of the genes described above. Point mutations can be introduced by a variety of methods including chemical mutagenesis, mutagenic copying methods and site specific mutagenesis methods using synthetic oligonucleotides.
Cassette mutagenesis methods are conveniently used to introduce point mutations into the specified regions of the PA or LF genes. A double-stranded oligonucleotide region containing alterations in the specified sequences of the gene is prepared. This oligonucleotide cassette region can be prepared by synthesizing an oligonucleotide with the sequence alteration in residues of the PA or LF gene, annealing to a primer, elongating with the large fragment of DNA polymerase and trimming with BstBI. This double-stranded oligonucleotide is ligated into the Bamhi/BstBI fragment from pYS5 and the PpuMI-BamHI fragment from pYS6 to produce an intact recombinant DNA. Other methods of producing the double stranded oligonucleotides and other recombinant DNA vectors can be practiced.
Chemical mutagenesis can be performed using the M13 vector system. A single strand M13 recombinant DNA is prepared containing recombinant PA or LF DNA. Another M13 recombinant containing the same recombinant DNA but in double stranded form is used to prepare a deletion in the targeted region of the gene. This double stranded M13 recombinant is cleaved into a linear molecule with an endonuclease, denatured, and annealed with the single strand M13 recombinant, resulting in a single strand gap in the target region of the PA or LF DNA.
This gapped DNA M13 recombinant is then treated with a compound such as sodium bisulfite to deaminate the cytosine residues in the single strand DNA region to uracil. This results in limited and specific mutations in the single strand DNA region. Finally, the gap in the DNA is filled in by incubation with DNA polymerase, resulting in a U-A base pair to replace a G-C base pair in the in unmutated portion of the gene. Upon replication the new recombinant gene contains T-A base pairs, which are point mutations from the original sequence. Other forms of chemical mutagenesis are also available.
Mutagenic copying of the PA or LF recombinant DNA can be carried out using several methods. For example, a single-stranded gapped DNA region is created as described above. This region is incubated with DNA polymerase I and one or more mutagenic analogs of normal ribonucleoside triphosphates. Copying of the single stranded region with the DNA polymerase substitutes the mutagenic analogs as the single strand gap region is filled in. Transfection and replication of the resulting DNA results in production of some mutated recombinant DNAs for PA, LF, or EF which can then be selected by cloning. Other mutagenic copying methods can be used.
Point mutations can be introduced into the specified regions of the PA or LF genes by methods using synthetic oligonucleotides for site-specific mutagenesis. PCR copying of the PA or LF genes is performed using oligonucleotide primers covering the specified target regions, and which contain modifications from the wild type sequence in these regions. The PA gene in a pYS5 vector can be PCR amplified using this method to result in mutations in the 164-167 position. PCR amplification can also be used to introduce mutations in the target region of the LF gene.
Synthetic oligonucleotide methods of introducing point mutations can be performed using heteroduplex DNA. A M13 recombinant DNA vector containing the PA or LF gene is prepared and a single-stranded M13 recombinant is produced. A single strand oligonucleotide containing an alteration in the specified target sequence for the PA or LF gene is annealed to the single strand M13 recombinant to produce a mismatched sequence. Incubation with DNA polymerase I results in a double-stranded M13 recombinant containing base pair mismatches in the specified region of the gene. This M13 recombinant is replicated in a host such as B. anthracis or E. coli to produce both wild type and mutant M13 recombinants. The mutated M13 recombinants are cloned and isolated. Other vector systems for mutagenesis involving synthetic nucleotides and heteroduplex formation can be applicable.
Expression of Proteins in Prokaryotic Cells
In addition to the use of cloning methods in bacteria such as Bacillus anthracis for amplification of cloned sequences, it may be desirable to express the proteins in other prokaryotes. It is possible to recover a functional protein from E. coli transformed with an expression plasmid encoding a PA or LF protein. Conveniently, the mutated PA proteins of the invention were expressed in B. anthracis and the LF-fusion proteins were expressed in E. coli.
Methods for the expression of cloned genes in bacteria are well known. See Sambrook. To optimize expression of a cloned gene in a prokaryotic system, expression vectors can be constructed which include a promoter to direct mRNA transcription termination. The inclusion of selection markers in DNA vectors transformed in bacteria are useful. Examples of such markers include the genes specifying resistance to ampicillin, tetracycline, or chloramphenicol.
See Sambrook, previously cited, for details concerning selection markers and promoters for use in bacteria such as E. coli. In an embodiment of this invention, pYS5 is a vector for the subcloning and amplification of desired gene sequences although other vectors could be used.
Strains of Bacillus Anthracis Producing Mutated Protein(s)
For PA protein production, B. anthracis strains cured of both pX01 and pX02 are preferred because they are avirulent. Examples of such strains are UM23Cl-1 and UM44-1C9, obtained from Curtis Thorne, University of Massachusetts. Similar strains can be made by curing of plasmids, as described by P. Mikesell, et al. , "Evidence for plasmid-mediated toxin production in Bacillus anthracis," Infect. Immun. 39:371-376 (1983).
See generally commonly assigned U.S. patent application Ser. No. 08/042,745, filed Apr. 5, 1993, now abandoned incorporated by reference herein.
Treatment Methods
A method for delivering a desired activity to a cell is provided. The steps of the method include administering to the cell (a) a protein comprising the translocation domain and the LF binding domain of the native PA protein and a ligand domain, and (b) a product comprising the PA binding domain of the native LF protein and a non-LF activity inducing moiety, whereby the product administered in step (b) is internalized into the cell and performs the activity within the cell.
The method of delivering an activity to a cell can use a ligand domain that is the receptor binding domain of the native PA protein. Other ligand domains are selected for their specificity for a particular cell type or class of cells. The specificity of the PA fusion protein for the targeted cell can be determined using standard methods and as described in Examples 2 and 3.
The method of delivering an activity to a cell can use an activity inducing moiety that is a polypeptide, for example a growth factor, a toxin, an antisense nucleic acid, or a nucleic acid encoding a desired gene product. The actual activity inducing moiety used will be selected based on its functional characteristics, e.g. its activity.
A method of killing a tumor cell in a subject is also provided. The steps of the method can include administering to the subject a first fusion protein comprising the translocation domain and LF binding domain of the native PA protein and a tumor cell specific ligand domain in an amount sufficient to bind to a tumor cell. A second fusion protein is also administered wherein the protein comprises the PA binding domain of the native LF protein and a cytotoxic domain of a non-LF protein in an amount sufficient to bind to the first protein, whereby the second protein is internalized into the tumor cell and kills the tumor cell.
The cytotoxic domain can be a toxin or it can be another moiety not strictly defined as a toxin, but which has an activity that results in cell death. These cytotoxic moieties can be selected using standard tests of cytotoxicity, such as the cell lysis and protein synthesis inhibition assays described in the examples.
The invention further provides a method of killing HIV-infected cells in a subject. The method comprises the steps of administering to the subject a first fusion protein comprising the translocation domain and LF binding domain of the native PA protein and a ligand domain that specifically binds to an HIV protein expressed on the surface of an HIV-infected cell, in an amount sufficient to bind to an HIV-infected cell. The next step is administering to the subject a second fusion protein comprising the PA binding domain of the native LF protein and a cytotoxic domain of a non-LF protein, in an amount sufficient to bind to the first protein, whereby the second protein is internalized into the HIV-infected cell and kills the HIV-infected cell, thereby preventing propagation of HIV.
Although certain of the methods of the invention have been described as using LF fusion proteins, it will be understood that other LF compositions having chemically attached activity inducing moieties can be used in the methods.
The fusion proteins and other compositions of the inventions can be administered by various methods, e.g., parenterally, intramuscularly or intrapertioneally.
The amount necessary can be deduced from other receptor/ligand or antibody/antigen therapies. The amount can be optimized by routine procedures. The exact amount of such LF and PA compositions required will vary from subject to subject, depending on the species, age, weight and general condition of the subject, the severity of the disease that is being treated, the particular fusion protein of composition used, its mode of administration, and the like. Generally, dosage will approximate that which is typical for the administration of cell surface receptor ligands, and will preferably be in the range of about 2 .mu.g/kg/day to 2 mg/kg/day.
Depending on the intended mode of administration, the compounds of the present invention can be in various pharmaceutical compositions. The compositions will include, as noted above, an effective amount of the selected protein in combination with a pharmaceutically acceptable carrier and, in addition, can include other medicinal agents, pharmaceutical agents, carriers, adjuvants, diluents, etc. By "pharmaceutically acceptable" is meant a material that is not biologically or otherwise undesirable, i.e., the material can be administered to an individual along with the fusion protein or other composition without causing any undesirable biological effects or interacting in a deleterious manner with any of the other components of the pharmaceutical composition in which it is contained.
Parenteral administration, if used, is generally characterized by injection. Injectables can be prepared in conventional forms, either as liquid solutions or suspensions, solid forms suitable for solution or suspension in liquid prior to injection, or as emulsions. A more recently revised approach for parenteral administration involves use of a slow release or sustained release system, such that a constant level of dosage is maintained. See, e.g., U.S. Pat. No. 3,710,795, which is incorporated by reference herein.
Formulations and Administration
Proteins of the invention such as PAHIV are typically mixed with a physiologically acceptable fluid prior to administration to a mammal such as a human. Examples of physiologically acceptable fluids include saline solutions such as normal saline, Ringer's solution, and generally mixtures of various salts including potassium and phosphate salts with or without sugar additives such as glucose. The proteins are administered parenterally with intravenous administration being the most typical route. Either a bolus of the protein in solution or a slow infusion can be administered intravenously. The choice of a bolus or an infusion depends on the kinetics, including the half-life, of the protein in the patient. An appropriate evaluation of the time for delivery of the protein is well within the skill of the clinician.
Patients selected for treatment with PAHIV are infected with HIV-1 and they may or may not be symptomatic. Optimally, the protein would be administered to an HIV-1 infected person who is not yet symptomatic. The dosage range of a protein of the invention such as PAHIV is typically from about 5 to about 25 micrograms per kilogram of body weight of the patient. Usually, the dose is about 10 micrograms per kilogram of body weight of the patient. The dosage is repeated at regular intervals, such as weekly for about 4 to 6 weeks. At that time the clinician may opt to evaluate the patient's immune status, including immuno-tolerance to the PAHIV, to decide future treatment.
The foregoing description and the following examples are offered primarily for purposes of illustration. It will be readily apparent to those skilled in the art that the operating conditions, materials, procedural steps and other parameters of the system described herein can be further modified or substituted in various ways without departing from the spirit and scope of the invention. For example, although human use has been discussed, veterinary use of the invention is also feasible. For instance, cats suffer from a so-called feline AIDS or feline immunodeficiency virus (FIV). Protective antigen can be altered to include a protease cleavage site specific for FIV. Thus, the invention is not limited by the description and examples, but rather by the appended claims.
EXAMPLE 1
Fusions of Anthrax Toxin Lethal Factor to the ADP-Ribosylation Domain of Pseudomonas Exotoxin
Reagents and General Procedures
Restriction endonucleases and DNA modifying enzymes were purchased from GIBCO/BRL, Boehringer Mannheim, or New England Biolabs. Low melting point agarose (Sea Plaque) was obtained from FMC Corp. (Rockland, Me.). Oligonucleotides were synthesized on a PCR Mate (Applied Biosystems) and purified on oligonucleotide purification cartridges (Applied Biosystems). The PCR was performed with a DNA amplification reagent (GeneAmp) from Perkin-Elmer Cetus Instruments and a thermal cycler (Perkin-Elmer Cetus). The amplification involved denaturation at 94.degree. C. for 1 min, annealing at 55.degree. C. for 2.5 min and extension at 72.degree. C. for 3 min, for 30 cycles. A final extension was run at 72.degree. C. for 7 min. For amplification of PE fragments, 10% formamide was added in the reaction mixture to decrease the effect of high GC content. DNA sequencing reactions were done using the Sequenase version 1.0 from U.S. Biochemical Corp. and DNA sequencing gels were made from Gel Mix 6 from GIBCO/BRL. .sup.35 S!deoxyadenosine 5'-.alpha.-thio!triphosphate and L-3,4,5-.sup.3 H!leucine were purchased from Dupont-New England Nuclear. J774A.1 cells were obtained from American Type Culture Collection. Chinese Hamster Ovary (CHO) cells were obtained from Michael Gottesman (National Cancer Institute, National Institutes of Health) (ATCC CCL 61).
Plasmid Construction
Construction of plasmids containing LF-PE fusions was performed as follows. Varying portions of the PE gene were amplified by PCR, ligated in frame to the 3' end of the LF gene, and inserted into the pVEX115 f+T expression vector (provided by V. K. Chaudhary, National Cancer Institute, National Institutes of Health). To construct fusion proteins, the 3'-end of the native LF gene (including codon 776 of the mature protein, specifying Ser) was ligated with the 5'-ends of sequences specifying varying portions of domains II, Ib, and III of PE. The LF gene was amplified from the plasmid pLF7 (Robertson, D. L. and Leppla, S. H. Gene 44:71-78, 1986) by PCR using PCR using oligonucleotide primers which added KpnI and MluI sites at the 5' and the 3' ends of the gene, respectively. Similarly, varying portions of the PE gene (provided by David FitzGerald, National Cancer Institute, National Institutes of Health) were amplified by PCR so as to add MluI and EcORI sites at the 5' and 3' ends. The PCR product of the LF gene was digested with KpnI and the DNA was precipitated. The LF gene was subsequently treated with MluI. Similarly, the PCR products of PE amplification were digested with MluI and EcoRI. The expression vector pVEX115 f+T was cleaved with KpnI and EcoRI separately and dephosphorylated. This vector has a T7 promoter, OmpA signal sequence, multiple cloning site, and T7 transcription terminator. All the above DNA fragments were purified from low-melting point agarose, a three-fragment ligation was carried out, and the product transformed into E. coli DH5.alpha. (ATCC 53868). The four constructs described in this report have the entire LF gene fused to varying portions of PE. The identity of each construct was confirmed by sequencing the junction point using a Sequenase kit (U.S. Biochemical Corp. ). For expression, recombinant plasmids were transformed into E. coli strain SA2821 (provided by Sankar Adhya, National Cancer Institute, National Institutes of Health, which is a derivative of BL21(.lambda.DE3) (Studier, F. W. and Moffatt, B. A. J. Mol. Biol. 189:113-150, 1986). This strain has the T7 RNA polymerase gene under control of an inducible lac promotor and also contains the degP mutation, which eliminates a major periplasmic protease (Strauch et al. J. Bacteriol. 171:2689-2696, 1989).
In the resulting plasmids, the LF-PE fusion genes are under control of the T7 promoter and contain an OmpA signal peptide to obtain secretion of the products to the periplasm so as to facilitate purification. The design of the PCR linkers also led to insertion of two non-native amino acids, Thr-Arg, at the LF-PE junction. The four fusions analyzed in this report contain the entire 776 amino acids of mature LF, the two added residues TR (Thr-Arg), and varying portions of PE. In fusion FP33, the carboxyl-terminal end of PE was changed from the native REDLK (Arg-Glu-Asp-Leu-Lys) (SEQ. ID NO: 33) to LDER (SEQ. ID NO: 34), a sequence that fails to cause retention in the ER (endoplasmic reticulum).
Expression and Purification of Fusion Proteins
Fusion proteins produced from pNA2, pNA4, pNA23 and pNA33 were designated FP2, FP4, FP23 and FP33 respectively. E. coli strains carrying the recombinant plasmids were grown in super broth (32 g/L Tryptone, 20 g/L yeast extract, 5 g/L NaCl, pH 7.5) with 100 .mu.g/ml of ampicillin with shaking at 225 rpm at 37.degree. C. in 2-L cultures. When A.sub.600 reached 0.8-1.0, isopropyl-1-thio-.beta.-D-galactopyranoside was added to a final concentration of 1 mM, and cultures were incubated an additional 2 hr. EDTA and 1,10-o-phenanthroline were added to 5 mM and 0.1 mM respectively, and the bacteria were harvested by centrifugation at 4000.times.g for 15 min at 4.degree. C. For extraction of the periplasmic contents, cells were suspended in 75 ml of 20% sucrose containing 30 mM Tris and 1 mM EDTA, incubated at 0.degree. for 10 min, and centrifuged at 8000.times.g for 15 min at 4.degree. C. Cells were resuspended gently in 50 ml of cold distilled water, kept on ice for 10 min, and the spheroplasts were pelleted. The supernatant was concentrated with CENTRI PREP.RTM.-100 units (Amicon) and loaded on a SEPHACRYL.RTM. S-200 (a cross-linked co-polymer of alkyl dextron and N,N'-methylenebisacrylanide column (40.times.2 cm) and 1 ml fractions were collected.
Fractions having full length fusion protein as determined by immunoblots were pooled and concentrated as above. Protein was then purified on an anion exchange column (Mono Q.RTM. HR5/5, Pharmacia-LKB) using a NaCl gradient. The fusion proteins eluted at 280-300 mM NaCl. The proteins were concentrated again on CENTRIPREP.RTM.-100 (Amicon Division) and the Mono Q.RTM. chromatography was repeated. Protein concentrations were determined by the bicinchoninic acid method (BCA Protein Assay Reagent, Pierce), using bovine serum albumin as the standard. Proteins were analyzed by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS). Gels were either stained with Coomassie Brilliant Blue or the proteins were electroblotted to nitrocellulose paper which was probed with polyclonal rabbit antisera to LF or PE (List Biological Laboratories, Campbell, Calif). To determine the percent of full length protein, SDS gels stained with Coomassie Brilliant Blue were scanned with a laser densitometer (Pharmacia-LKB Ultrascan XL).
The proteins migrated during gel electrophoresis with molecular masses of more than 106 kDa, consistent with the expected sizes, and immunoblots confirmed that the products had reactivity with antisera to both LF and PE. The fusion proteins differed in their susceptibility to proteolysis as judged by the appearance of smaller fragments on immunoblots, and this led to varying yields of final product. Thus, from 2-L cultures the yields were FP2, 27 .mu.g; FP4, 87 .mu.g; FP23, 18 .mu.g; and FP33, 143 .mu.g.
Cell Culture Techniques and Protein Synthesis Inhibition Assay
CHO cells were maintained as monolayers in Eagle's minimum essential medium (EMEM) supplemented with 10% fetal bovine serum, 10 mM 4-2(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (pH 7.3), 2 mM glutamine, penicillin/streptomycin, and non-essential amino acids (GIBCO/BRL). Cells were plated in 24- or 48-well dishes one day before the experiment. After overnight incubation, the medium was replaced with fresh medium containing 1 .mu.g/ml of PA unless otherwise indicated. Fusion proteins were added to 0.1-1000 ng/ml. All data points were done in duplicate. Cells were further incubated for 20 hr at 37.degree. C. in 5% CO.sub.2 atmosphere. The medium was then aspirated and cells were incubated for 2 hr at 37.degree. C. with leucine-free medium containing 1 .mu.Ci/ml .sup.3 H!leucine. Cells were washed twice with medium, cold 10% trichloroacetic acid was added for 30 min, the cells were washed twice with 5% trichloroacetic acid and dissolved in 0.150 ml 0.1M NaOH. Samples were counted in Pharmacia-LKB 1410 liquid scintillation counter. In experiments to determine if the toxin is internalized through acidified endosomes, 1 .mu.M monensin (Sigma) was added 90 min prior to toxin and was present during all subsequent steps. To verify that the fusion proteins were internalized through the PA receptor, competition with native LF was carried out. PA (0.1 .mu.g/ml) and LF (0.1-10,000 ng/ml) were added to the CHO cells to block the PA receptor and the fusion proteins were added thereafter at concentrations of 100 ng/ml for FP4 and FP23 and 5 ng/ml for FP 33. Protein synthesis inhibition was measured after 20 hr as described above.
Cytotoxic Activity of the Fusion Proteins
All four fusion proteins made and purified were toxic to CHO cells. The concentration causing 50% lysis of cultured cells (EC.sub.50) values of the proteins were 350, 8, 10, and 0.2 ng/ml for FP2, FP4, FP23 and FP33 respectively (Table 1). These assays were done with PA present at 1 ug/ml, exceeding the K.sub.m of 0.1 ug/ml (100 pM). The fusion proteins had no toxicity even at 1 .mu.g/ml when PA was omitted, proving that internalization of the fusion proteins was occurring through the action of PA and the PA receptor. Native LF has previously been shown to have no short-term toxic effects on CHO cells when added with PA, and therefore was not included in these assays. The fusion protein having only domain III and an altered carboxyl-terminus (FP33) was most active, whereas the one having the intact domains II and III and the native REDLK (SEQ. ID NO: 33) terminus (FP2) was least active. The other two fusion proteins (FP4 and FP23) had intermediate potencies.
Among proteins having ADP-ribosylation activity, potencies equalling or exceeding 1 pM have previously been found only for native diphtheria and Pseudomonas toxins acting on selected cells (Middlebrook, J. L. and Dorlan, R. B. Can. J. Microbiol. 23:183-189, 1977) and for fusion proteins of PE and diphtheria toxin when tested on cells containing>100,000 receptors for the ligand-recognition domain of the fusion (EGF, transferrin, etc. ) (Pastan, I. and FitzGerald, D. Science 254:1173-1177, 1991; Middlebrook, et al. 1977). For CHO cells, the potency of FP33 (EC.sub.50 =2 pM) is higher than that of PE itself (EC.sub.50 =420 pM), even though CHO cells probably have similar numbers of receptors for both PA and PE (approx. 5,000-20,000). If the intracellular trafficking of native PE delivers less than 5% of the molecules to the cytosol, then the 200-fold greater potency of FP33 suggests that the PA/LF system has an inherently high efficiency of delivery to the cytosol.
A comparison of the potencies of the four fusion proteins shows that inclusion of domain II decreases potency. Thus, the fusion with the lowest potency, FP2, was the one containing intact domains II, Ib, and III. In designing the fusion proteins, all or part of PE domain II and Ib was included in several of the constructs because it could not be assumed that the translocation functions possessed by PA and LF would be able to correctly traffic PE domain III to the cytosol. The combination of domains II, Ib, and III, termed PE40, has been used in a large number of toxic hybrid proteins, by fusion to growth factors, monoclonal antibodies, and other proteins (Pastan et al. 1991; Oeltmann, T. N. and Frankel, A. E. FASEB J. 5:2334-2337, 1991), and some of these fusions have shown substantial potency. Domain II was found to be essential in these hybrid proteins to provide a translocation function not present in the receptor-binding domain to which it was fused. The potency of many of these PE40 fusion proteins appears to require that they be trafficked through the Golgi and ER and proteolytically activated in the same manner as native PE, so as to achieve delivery of domain III to the cytosol. The fact that inclusion of the entire domain II in the LF fusion protein FP2 instead decreased activity suggests that internalization of the LF fusions occurs through a different route, one that does not easily accommodate all the sequences in domain II.
Evidence that structures within PE residues 251-278 inhibit translocation of the LF fusions comes from the 35-fold lower potency of FP2 compared to FP23. One structure that might inhibit translocation of the fusions is the disulfide loop formed by Cys265 and Cys287. In native PE, this disulfide loop appears to be required for maximum activity. Thus, native PE and TGF-.alpha.-PE40 fusions become 10- to 100-fold less toxic if one or both these cysteines are changed to serine. The disulfide loop probably acts to constrain the polypeptide so that Arg276 and Arg279 are susceptible to the intracellular protease involved in the cleavage that precedes translocation. In contrast, the disulfide loop decreases the potency of the LF fusions, perhaps by preventing the unfolding needed for passage through a protein channel, thereby acting in this situation as a "stop transfer" sequence. FP23, which lacks Cys265, would not contain the domain II disulfide, and therefore would not be subject to this effect. LF, like PA and EF, contains no cysteines, and would not be prevented by disulfide loops from the complete unfolding needed to pass through a protein channel. The suggestion that disulfide loops act as stop-transfer signals would predict that the disulfide Cys372-Cys379 in PE domain Ib, which is retained in all four LF fusions would also decrease potency. It should be noted that neither the fusions made here nor the PE40 fusions have been analyzed chemically to determine if the disulfides in domains II and III are actually formed. If the disulfides do form correctly, it would be predicted that the potencies of all of the fusion proteins, and especially that of FP2, would be increased by treatment with reducing agents. These analyses have not yet been performed. This analysis also suggests that future LF fusions might be made more potent by omission of domain Ib.
The other structural feature of PE known to affect intracellular trafficking is the carboxyl terminal sequence, REDLK (SEQ. ID NO: 33), that specifies retention in the ER (Chaudhary et al. 1990; Muro et al. 1987). To determine if the trafficking of the LF fusion proteins was similar to that of PE, two of the fusion proteins were designed so as to differ only in the terminal sequence. Replacement of the native sequence by LDER (SEQ. ID NO: 34), one that does not function as an ER retention signal, produced the most toxic of the four fusion proteins, FP33. FP4, identical except that it retained a functional REDLK (SEQ. ID NO: 33) sequence, was 30-fold less potent. These data suggest that sequestration of the REDLK-ended (SEQ. ID NO: 33) fusions decreased their access to cytosolic EF-2. The implication is that PE may require the REDLK (SEQ. ID NO: 33) terminus to be delivered to the ER for an obligatory processing step, but then be limited in its final toxic potential by sequestration from its cytosolic target. Finally, this comparison strongly argues that internalization of the LF fusions does not follow the same path as PE.
In designing the fusion proteins described here it was hoped that they would have cytotoxic activity against cells that are unaffected by anthrax lethal toxin, and this was successfully realized as shown by the data obtained with CHO cells. However, prior knowledge about LF did not provide a basis for predicting whether the constructs would retain toxicity toward mouse macrophages, the only cells known to be rapidly killed by anthrax lethal toxin. Macrophages are lysed by lethal toxin in 90-120 minutes, long before any inhibition of protein synthesis resulting from ADP-ribosylation of EF-2 leads to decreases in membrane integrity or viability. This kinetic difference made it possible to test directly for LF action. As discussed above, the fusion proteins purified to remove the -89-kDa LF species formed by proteolysis were not toxic to J774A.1 macrophages. This shows that attachment of a bulky group to the carboxyl terminus of LF eliminates its normal toxic activity. In the absence of any assay for the putative catalytic activity of LF, it is not possible to determine the cause of the loss of LF activity. The inability of the fusions to lyse J774A.1 cells also argues against proteolytic degradation of the fusions either in the medium during incubation with cells or after internalization.
An important result of the invention described here is the demonstration that the anthrax toxin proteins constitute an efficient mechanism for protein internalization into animal cells. The high potency of the present fusion proteins argues that this system is inherently efficient, as well as being amenable to improvement. The high efficiency results in part from the apparent direct translocation from the endosome, without a requirement for trafficking through other intracellular compartments. In addition to its efficiency, the system appears able to tolerate heterologous polypeptides.
Macrophage Lysis Assay of Fusion Proteins
Fusion proteins were assayed for LF functional activity on J774A.1 macrophage cell line in the presence of 1 .mu.g/ml PA. One day prior to use, cells were scraped from flasks and plated in 48-well tissue culture dishes. For cytotoxicity tests, the medium was aspirated and replaced with fresh medium containing 1 .mu.g/ml PA and the LF fusion proteins, and the cells were incubated for 3 hr. All data points were performed in duplicate. To measure the viability of the treated cells, 3-4,5-dimethylthiazol-2-yl!-2,5-diphenyltetrazolium bromide (MTT) was added to the cells to a final concentration of 0.5 mg/ml, and incubation was continued for an additional 45 min to allow the uptake and oxidation of MTT by viable cells. Medium was aspirated and replaced by 200 .mu.l of 0.5% SDS, 40 mM HCl, 90% isopropanol and the plates were vortexed to dissolve the blue pigment. The MTT absorption was read at 570 nm using a UVmax Kinetic Microplate Reader (Molecular Devices Corp.).
The crude periplasmic extracts from which the fusion proteins were purified caused lysis of J774A.1 macrophages when added with PA, indicating the presence of active LF species, probably formed by proteolysis of the fusion proteins. Purification removed this activity, so that none of the final fusion proteins had this activity. This result showed both that the purified proteins were devoid of full size LF or active LF fragments, and that the lytic activity of LF for macrophages is blocked when residues from PE are fused at its carboxyl terminus.
ADP-Ribosylation Assays
For assaying ADP-ribosylation activity, the method of Collier and Kandel (Collier, R. J. and Kandel, J. J. Biol. Chem. 246:1496-1503, 1971) was used with some modification. A wheat germ extract enriched for EF-2 was used in the reaction. Briefly, in a 200-.mu.L reaction assay, 20 .mu.L of buffer (500 mM Tris, 10 mM EDTA, 50 mM dithiothreitol and 10 mg/ml bovine serum albumin) was mixed with 30 .mu.L of EF-2, 130 .mu.L of H.sub.2 O or sample, and 20 .mu.L of adenylate-.sup.32 P!NAD (0.4 .mu.Ci per assay, ICN Biochemicals) containing 5 .mu.M of non-radioactive NAD. Samples were incubated for 20 min at 23.degree. C., the reactions were stopped by adding 1 ml 10% trichloroacetic acid, and the precipitates were collected and washed on GA-6 filters (Gelman Sciences). The filters were washed twice with 70% ethanol, air dried, and the radioactivity measured.
Table 1 shows that all the fusion proteins were equally capable of ADP-ribosylation of EF-2. FP2, which had little cytotoxic activity on CHO cells, still retained full ADP-ribosylation activity. It was also found that treatment with urea and dithiothreitol under conditions that activate the enzymatic activity of native PE, caused no increase in the ADP-ribosylation activity of the fusion proteins, suggesting that the proteins were not folded so as to sterically block the catalytic site.
Effect of Mutant PA on LF-PE Activity
To verify that uptake of the fusion proteins requires PA, the activity of the fusion proteins was measured in the presence of a mutant PA which is apparently defective in internalization. This mutant, PA-S395C, has a serine to cysteine substitution at residue 395 of the mature protein, and retains the ability to bind to receptor, become proteolytically nicked, and bind LF, but is unable to lyse macrophages. When PA-S395C was substituted for native PA in combination with FP33, no inhibition of protein synthesis inhibition was observed. Similar results were obtained when the other three fusion proteins were tested in combination with PA-S395C.
Effect of Monensin on Activity of the Fusion Proteins
To verify that internalization of the fusion proteins was occurring by passage through acidified endosomes in the same manner as native LF, the ability of monensin to protect cells was examined. Addition of monensin to 1 .mu.M decreased the potency of FP33 by >100-fold. Protection against the other three fusion proteins exceeded 20-fold.
LF Block of LF-PE Fusion Activity
To further verify that the fusion proteins were internalized through the PA receptor, CHO cells were incubated with PA and different amounts of LF to block the receptor and the fusion proteins were added thereafter. Protein synthesis inhibition assays showed that native LF could competitively block LF-PE fusion proteins in a concentration-dependent manner.
The present data suggest that the receptor-bound 63-kDa proteolytic fragment of PA forms a membrane channel and that regions at or near the amino-termini of LF and EF enter this channel first and thereby cross the endosomal membrane, followed by unfolding and transit of the entire polypeptide to the cytosol. This model differs from that for diphtheria toxin in that the orientation of polypeptide transfer is reversed. Since both EF and LF have large catalytic domains, extending to near their carboxyl termini, it appears probable that the entire polypeptide crosses the membrane. In the LF fusion proteins, the attached PE sequences would be carried along with the LF polypeptide in transiting the channel to the cytosol. Thus, the PA63 protein channel must tolerate diverse amino acid residues and sequences. The data presented is consistent with the mechanism of direct translocation of the LF proteins to the cytosol as suggested herein.
EXAMPLE 2
Residues 1-254 of Anthrax Toxin Lethal Factor are Sufficient to Cause Cellular Uptake of Fused Polypeptides Reagents and General Procedures
Restriction endonucleases and DNA modifying enzymes were purchased from GIBCO/BRL, Boehringer Mannheim or New England Biolabs. Low melting point agarose (Sea Plaque) was obtained from FMC Corporation. Oligonucleotides were synthesized on a PCR Mate (Applied Biosystems) and purified with Oligonucleotide Purification Cartridges (Applied Biosystems). Polymerase chain reactions (PCR) were performed on a thermal cycler (Perkin-Elmer-Cetus) using reagents from U. S. Biochemical Corp. or Perkin-Elmer-Cetus. DNA was amplified as described in Example 1. The DNA was sequenced to confirmed the accuracy of all of the constructs described in the report. SEQUENASE version 2.0 from U.S. Biochemical Corp. was utilized for the sequencing reactions, and DNA sequencing gels were made with Gel Mix 8 from GIBCO/BRL. .sup.35 S!dATP.alpha.S and L-3,4,5-.sup.3 H!leucine were purchased from Dupont-New England Nuclear. Chinese hamster ovary cells (CHO) were obtained from Michael Gottesman (NCI, NIH). J774A.1 macrophage cells were obtained from American Type Culture Collection.
Plasmid Construction
Three types of LF protein constructs were made and analyzed in this report. All the constructs were made by PCR amplification of the desired sequences, using the native LF gene as template. LF proteins deleted at the amino- or carboxyl-terminus were constructed by a single PCR amplification reaction that added restriction sites at the ends for incorporation of the construct into the expression vector. LF proteins deleted for one or more of the 19-amino acid repeats that comprise residues 308-383 were constructed by ligating the products of two separate PCR reactions that amplified the regions bracketing the deletion. The third group of constructs were fusions of varying portions of the amino terminus of LF with PE domains Ib and III. Like the internally-deleted LF proteins, these LF-PE fusions were also made by ligation of two separate PCR products. In the latter two types of constructs, the ligation of the PCR products resulted in addition of a linker, ACGCGT, at the junction points. This introduced two non-native residues, Thr-Arg, between the fused domains. The PCR manipulations also added three non-native amino acids, Met-Val-Pro, as an extension to the native amino terminus on all the constructs described in this report. Addition of this sequence is not likely to alter the activity of the constructs (discussed below). It should be noted that the LF-PE fusions described herein contain this three-residue extension.
For PCR reactions to make deletions of 40 and 78 amino acids from the amino-terminus of LF, two different mutagenic oligonucleotide primers were made which were substantially identical to the LF gene template at the intended new termini, and which added KpnI sites at their 5'-ends. Another (non-mutagenic) oligonucleotide primer for introduction of a BamHI site at the 3' end of LF was prepared. Similarly, to make deletions at the carboxyl-terminus of LF, two different mutagenic primers were used which truncated LF at residues 729 and 693 and introduced a BamHI site next to the new 3' ends of the LF gene. A second (non-mutagenic) oligonucleotide primer specific for the amino terminus of LF was made which introduced a KpnI site at the 5' end of the gene. All of the primers noted above were used in PCR reactions on a pLF7 template (Robertson and Leppla, 1986) to synthesize DNA fragments having KpnI and BamHI sites at their 5' and 3' ends, respectively. The amplified LF DNAs containing the amino- and carboxyl-terminal deletions were digested with the appropriate restriction enzymes. The expression vector pVEX115f+T (provided by V. K. Chaudhary, NCI, NIH) was cleaved sequentially with KpnI and BamHI and dephosphorylated. This expression vector contains a T7 promoter, an OmpA signal sequence for protein transport to the periplasm, a multiple cloning site that includes KpnI and BamHI sites, and a T7 transcription terminator. The LF and pVEX115f+T DNA fragments were purified from low melting point agarose, ligated overnight, and transformed into E. coli DH5.alpha.. Transformants were screened by restriction digestion to identify the desired recombinant plasmids. Proteins produced by these constructs are designated according to the amino acid residues retained; for example the LF truncated at residue 693 is designated LF.sup.1-693. All of the mutant LF proteins described above contain three non-native amino acids, Met-Val-Pro, added to the amino-terminus as a result of the PCR manipulations.
To analyze the role of the repeat region of LF, four different constructs were made: 1., removal of the entire repeat region (LF.sup.1-307.TR.LF.sup.384-776), 2., removal of the first repeat (LF.sup.1-307.TR.LF.sup.327-776) 3., removal of the last repeat (LF.sup.1-364.TR.LF.sup.384-776), and 4., removal of repeats 2-4 (LF.sup.1-326.TR.LF.sup.384-776). To construct LF.sup.1-307.TR.LF.sup.384-776, four different primers were used in two separate PCR reactions. To amplify LF.sup.1-307, one oligonucleotide primer was made at the 5'-end of the LF gene which added a KpnI site, and a second primer was constructed at the end of residue 307, introducing an MluI site. For amplifying LF.sup.384-776, a third primer was made at residue 384 with an added MluI site, and the fourth primer was made at the residue 776 which introduced a BamHI site at the end. Two PCR amplifications were done using primers one/two and three/four with pLF7 as template (Robertson and Leppla, 1986). The first amplification reaction was digested with KpnI and MluI separately, and the second amplification reaction was digested with MluI and BamHI. The expression vector pVEX115f+T was digested separately with KpnI and BamHI and dephosphorylated. All three fragments were gel purified, ligated overnight at 16.degree. C. and transformed into E. coli DH5.alpha.. The other three constructs were made by similar strategies. Oligonucleotide primers one and four were the same for all four constructs, whereas primers two and three were changed accordingly. All four constructs contain Met-Val-Pro at the amino terminus of LF and Thr-Arg at the site of the repeat region deletion.
To construct LF-PE fusion proteins, fragments of the LF gene extending from the amino terminus to various lengths were amplified from plasmid pLF7 (Robertson and Leppla, 1986) by PCR using a common oligonucleotide primer that added a KpnI site at the 5' end and mutagenic primers which added MluI sites at the intended new 3' ends. The PCR products of the LF gene were digested with KpnI, the DNAs were precipitated, and subsequently digested with MluI. Domains Ib and III of the PE gene (provided by David FitzGerald, NCI, NIH) were amplified by PCR using primers which added MluI and EcoRI sites at the 5' and 3' ends, respectively. The PCR product of PE was digested with MluI and EcoRI. Similarly, the expression vector pVEX115f+T was digested with KpnI and EcoRI. All DNA fragments were purified from low-melting agarose gels, three-fragment ligations were carried out, and the products were transformed into E. coli DH5.alpha.. The three constructs described in this example have 254, 198 and 79 amino acids of LF joined with PE domains Ib and III. These fusion proteins are designated LF.sup.1-254.TR.PE.sup.362-613 (SEQ. ID NO: 10), LF.sup.1-198.TR.PE.sup.362-613, and LF.sup.1-79.TR.PE.sup.362-613, respectively. The proteins retain the native carboxyl-terminal sequence of PE, REDLK (SEQ. ID NO: 33). It should be noted that these abbreviations do not specify the entire amino acid content of the proteins, because all the constructs also contain Met-Val-Pro, which was added to the amino-terminus of the LF domain by the PCR manipulations.
Expression and Purification of Deleted LF and Fusion Proteins
Recombinant plasmids were transformed into E. coli SA2821 (provided by Sankar Adhya, NCI, NIH), a derivative of BL21(.lambda.DE3) (Studier and Moffatt, 1986) that lacks the proteases encoded by the lon, OmpT, and degP genes, and has the T7 RNA polymerase gene under control of the lac promoter (Strauch et al., 1989). Transformants were grown in super broth with 100 .mu.g/ml ampicillin, with shaking at 225 rpm, 37.degree. C., in 2-L cultures. When A.sub.600 reached 0.8-1.0, isopropyl-1-thio-.beta.-D-galactopyranoside was added to a final concentration of 1 mM, and cultures were incubated for an additional 2 h. EDTA and 1,10-o-phenanthroline were added to 5 and 0.1 mM, respectively, and periplasmic protein was extracted as described in Example 1. The supernatant fluids were concentrated by CENTRIPREP.RTM.-30 units (Amicon) and proteins were purified to near homogeneity by gel filtration SEPHACRYL.RTM. S-200 (a cross-linked co-polymer of alkyl dextron and N,N,'-methylenebisacrylanide Pharmacia-LKB) and anion exchange chromatography (Mono Q.RTM., Pharmacia-LKB) as described in Example 1. To determine the percentage of full length protein, SDS gels stained with Coomassie Brilliant Blue were scanned with a laser densitometer (Pharmacia-LKB Ultrascan XL). Western blots were performed as described previously (Singh et al., 1991).
The LF proteins having terminal deletions and the LF-PE fusion proteins were obtained from periplasmic extracts and purified to near homogeneity by gel filtration and anion exchange chromatography. The migration of the proteins was consistent with their expected molecular weights. Immunoblots confirmed that the LF proteins had reactivity with LF antisera, and the LF-PE fusion proteins had reactivity with both LF and PE antisera. Fusion proteins and terminally-deleted LF proteins differed in their susceptibility to proteolysis as judged by the appearance of peptide fragments on the immunoblots, and this was also reflected in the different amounts of purified proteins obtained. Thus, from 2-L cultures the yields of purified proteins were LF.sup.41-776, 39 .mu.g; LF.sup.79-776, 32 .mu.g; LF.sup.1-729, 50 .mu.g; LF.sup.1-693, 46 .mu.g; LF.sup.1-254.TR.PE.sup.362-613, 184 .mu.g; LF.sup.1-198.TR.PE.sup.362-613, 80 .mu.g; LF.sup.1-79.TR.PE.sup.362-613, 127 .mu.g.
LF proteins deleted in the repeat region were found to be unstable and full size product could not be purified. Therefore, the activities of these proteins were determined by assay of crude periplasmic extracts, and immunoblots were used to estimate the amount of the full size proteins present.
Cytotoxicity on Macrophages of LF Proteins Having Terminal and Internal Deletions
Deleted LF proteins were assayed for LF functional activity on the J774A.1 macrophage cell line in the presence of native PA as described in Example 1. Briefly, cells were plated in 24- or 48-well dishes in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum, and allowed to grow for 18 h. PA (1 .mu.g/ml) and the mutant LF proteins were added and cells were incubated for 3 h. To measure the viability of the treated cells, 3-4,5-dimethylthiazol-2-yl!-2,5-diphenyltetrazolium bromide (MTT) was added to the cells to a final concentration of 0.5 mg/ml. After incubating for 45 min, the medium was aspirated and cells were dissolved in 90% isopropanol, 0.5% SDS, 40 mM HCl, and read at 540 nm using a UVmax Kinetic Microplate Reader (Molecular Devices Corp.).
To determine the extent of essential sequences at the amino terminus of LF, the toxicities of the two LF proteins deleted at the amino-terminus were measured in combination with PA in the macrophage lysis assay. Purified LF.sup.41-776 and LF.sup.79-776 were unable to lyse J774A.1 macrophage cells. This indicates that some portion of the sequence preceding residue 41 is needed to maintain an active LF protein.
To examine the role of the carboxyl terminus of LF, two proteins truncated in this region were prepared and analyzed. The proteins LF.sup.1-693 and LF.sup.1-729 were assayed on J774A.1 cells and found to be inactive. This is presumed to be due to inactivation of the putative catalytic domain.
To begin study of the role of the repeat region of LF, four constructs were made having deletions in this region. The proteins expressed from these mutants were unstable. Of the four deleted proteins, only LF.sup.1-307.TR.LF.sup.327-776 had immunoreactive material at the position expected of intact fusion protein. The amount of intact LF.sup.1-307.TR.LF.sup.327-776 was similar to that of native LF expressed in the same vector. When these unpurified periplasmic extracts were tested in J774A.1 macrophages, only the native LF control was toxic. LF.sup.1-307.TR.LF.sup.327-776 did not lyse macrophages even when present at 50-fold higher concentration than that of crude periplasmic protein of LF. Conclusions cannot be drawn about the toxicities of the other three constructs because full size fusion proteins were not present in the periplasmic extracts.
Cell Culture Techniques and Protein Synthesis Inhibition Assay of Fusion Proteins
CHO cells were maintained as monolayers in .alpha.-modified minimum essential medium (.alpha.-MEM) supplemented with 5% fetal bovine serum, 10 mM HEPES (pH 7.3), and penicillin/streptomycin. Protein synthesis assays were carried out in 24- or 48-well dishes as described in Example 1. CHO cells were incubated with PA (0.1 ug/ml) and varying concentrations of LF, which is expected to block the receptor. Fusion proteins were added at fixed concentrations, as follows: FP4, 100 ng/ml, FP23, 100 ng/ml, and FP33, 5 ng/ml. Cells were incubated for 20 hr and protein synthesis inhibition was evaluated by .sup.3 H!leucine incorporation.
Cytotoxicity of the LF-PE Fusion Proteins on CHO Cells
The use of fusion proteins provides a more defined method for measuring the translocation of LF, as demonstrated in Example 1 showing that fusions of LF with domains Ib and III of PE are highly toxicy. Translocation of these fusions is conveniently measured because domain III blocks protein synthesis by ADP-ribosylation of elongation factor 2. The new fusions containing varying portions of LF fused to PE domains Ib and III were designed to identify the minimum LF sequence able to promote translocation. The EC.sub.50 of LF.sup.1-254.TR.PE.sup.362-613 (SEQ. ID NO: 10) was 1.7 ng/ml, whereas LF.sup.1-198.TR.PE.sup.362-613 and LF.sup.1-79.TR.PE.sup.362-613 did not kill 50% of the cells even at a 1200-fold higher concentration. Other constructs were also made and analyzed, containing larger portions of LF fused to PE domains Ib and III, and found those to be equal in potency to LF.sup.1-254.TR.PE.sup.362-613. These results show that residues 1-254 contain all the sequences essential for binding to PA63. The fusion proteins had no toxicity in the absence of PA, proving that their internalization absolutely requires interaction with PA.
Binding of Fusion Proteins and Deleted LF Proteins to PA
Binding of LF proteins to cell bound PA was determined by competition with radiolabeled .sup.125 I-LF. Native LF was radiolabeled (3.1.times.10.sup.6 cpm/.mu.g protein) using the Bolton-Hunter reagent. Binding studies employed the L6 rat myoblast cell line, which has approximately twice as many receptors as the J774A.1 macrophage line (Singh et al., 1989). For convenience, cells were chemically fixed by a gentle procedure that preserves the binding activity of the receptor as well as the ability of the cell-surface protease to cleave PA to produce receptor-bound PA63. Assays were carried out in 24-well dishes using cells plated in DMEM with 10% fetal bovine serum one day before the experiment. Cell monolayers were washed twice with Hanks' balanced salt solution (HBSS) containing 25 mM HEPES and were chemically fixed for 30 min at 23.degree. in 10 mM N-hydroxysuccinimide and 30 mM 1-ethyl-3-3-dimethylaminopropyl! carbodiimide, in buffer containing 10 mM HEPES, 140 mM NaCl, 1 mM CaCl.sub.2, and 1 mM MgCl.sub.2. Monolayers were washed with HBSS containing 25 mM HEPES and the fixative was inactivated by incubating 30 min at 23.degree. in DMEM (without serum) containing 25 mM HEPES. Native PA was added at 1 .mu.g/ml in minimum essential medium containing Hanks' salts, 25 mM HEPES, 1% bovine serum albumin, and a total of 4.5 mM NaHCO.sub.3. Cells were incubated overnight at room temperature to allow binding and cleavage of PA. Cells were washed twice in HBSS and mutant LF proteins (0-5000 ng/ml) along with 50 ng/ml .sup.125 I-LF was added to each well. Cells were further incubated for 5 h, washed three times in HBSS, dissolved in 0.5 ml 1N NaOH, and counted in a gamma counter (Beckman Gamma 9000).
Using this assay, the LF mutant proteins having amino-terminal deletions were found incapable of binding to PA, thereby explaining their lack of toxicity. Carboxyl-terminal deleted LF proteins did bind to PA in a dose dependent manner, although they had slightly lower affinity than LF. The proteins deleted in the repeat region could not be tested for competitive binding because their instability prevented purification of intact protein.
The EC.sub.50 for LF.sup.1-254.TR.PE.sup.362-613 binding was found to be 220 ng/ml, which is similar to that of LF, 300 ng/ml. Therefore the binding data correlate well with the toxicity of this construct. In contrast, neither LF.sup.1-198.TR.PE.sup.362-613 nor LF.sup.1-79.TR.PE.sup.362-613 bound to PA63 on cells, thereby explaining their lack of toxicity.
EXAMPLE 3
Construction of Genes Encoding PA Fusion Proteins
The genes encoding PA (or PA truncated at the carboxyl terminus to abrogate binding to the PA receptor) and an alternative targeting moiety (a single-chain antibody, growth factor, or other cell type-specific domain) are spliced using conventional molecular biological techniques. The PA gene is readily available, and the genes encoding alternative targeting domains are derived as described below.
Single-chain antibodies (sFv)
See Example 4, below.
Growth Factors and Other Targeting Proteins
The nucleotide sequences of genes encoding a number of growth factors and other proteins that are targeted to specific cell types or classes are reported in freely accessible databases (e.g., GenBank), and in many cases the genes are available. In circumstances where this is not the case, genes can be cloned from genomic or cDNA libraries, using probes based on the known nucleotide sequence of the gene that codes for the growth factor, or derived from a partial amino acid sequence of the protein (see, e.g. Sambrook, supra. ). Alternatively, genes encoding the growth factor or other targeting moiety can be produced de novo from chemically synthesized overlapping oligonucleotides, using the preferred codon usage of the expression host. For example, the gene for human epidermal growth factor urogastrone was synthesized from the known amino acid sequence of human urogastrone using yeast preferred codons. The cloned DNA, under control of the yeast GAPDH promoter and yeast ADH-1 terminator, expresses a product having the same properties as natural human urogastrone. The product of this synthesized gene is nearly identical to that of the natural urogastrone, the only difference being that the product of the synthetic gene has a trptophan at amino acid 13, while the other has a tyrosine (Urdea et al. Proc. Natl. Acad. Sci. USA 80:7461-7465, 1983).
Expression of PA Fusion Proteins
Once constructed, genes encoding PA-fusion proteins are expressed in Bacillus anthracis, and recombinant proteins are purified by one of the following methods: (i) size-based chromatographic separation; (ii) affinity chromatography. In the case of PA-sFv fusions, immobilized metal chelate affinity chromatography may be the purification method of choice, because addition of a string of six histidine residues at the carboxyl terminus of the sFv will have no detrimental effect on binding to antigen. Additional methods of expression of PA-fusion proteins utilize an in vitro rabbit reticulocyte lysate-based coupled transcription/translation system, which has been demonstrated to accurately refold chimeric proteins consisting of an sFv fused to diphtheria toxin, or Pseudomonas exotoxin A as demonstrated in Example 4.
Functional Testing of PA Fusion Proteins
After expression and purification, functionality of PA-fusion proteins are tested by determining their ability to act in concert with an LF-PE fusion protein to inhibit protein synthesis in an appropriate cell line. Using a PA-anti human transferrin receptor sFv fusion as a model, the following properties are examined: (i) Cell type-specificity (protein synthesis should be inhibited in cell lines which express the human transferrin receptor, but not in those which do not); (ii) Independence of toxicity from PA receptor binding (excess free PA should have no effect on toxicity of the PA-sFv/LF-PE complex); (iii) Competitive inhibition by excess free antibody (toxicity should be abrogated in the presence of excess sFv, or the monoclonal antibody from which it was derived). For example such tests are described in Examples 4 and 5. These studies and other studies are used to confirm that PA has been successfully re-routed to an alternative receptor to permit the use of the present anthrax toxin-based cell type-specific cytotoxic agents for the treatment of disease.
EXAMPLE 4
Generating Fusion Proteins with Single-Chain Antibodies Reagents
Methionine-free rabbit reticulocyte lysate-based coupled transcription/translation reagents, recombinant ribonuclease inhibitor (rRNasin), and cartridges for the purification of plasmid DNA were purchased from Promega (Madison, Wis.). Tissue culture supplies were from GIBCO (Grand Island, N.Y.) and Biofluids (Rockville, Md.). OKT9 monoclonal antibody was purchased from Ortho Diagnostic Systems (Raritan, N.J.). PCR reagents were obtained from by Perkin-Elmer Cetus Instruments (Norwalk, Conn.), and restriction and nucleic acid modifying enzymes (including M-MLV reverse transcriptase) were from GIBCO-BRL (Gaithersburg, Md.). A Geneclean kit for the recovery of DNA from agarose gels was supplied by BIO 101 (La Jolla, Calif.). Hybridoma mRNA was isolated using a Fast Trak mRNA isolation kit (Invitrogen, San Diego, Calif.). All isotopes were purchased from Du Pont-New England Nuclear (Boston, Mass.), except Adenylate-.sup.32 P!NAD, which was supplied by ICN Biomedicals (Costa Mesa, Calif.). Pseudomonas exotoxin A was obtained from List Biologicals (Campbell, Calif.). Oligonucleotides were synthesized on a dual column Milligen-Biosearch Cyclone Plus DNA synthesizer (Burlington, Mass.), and purified using OPC cartridges (Applied Biosystems, Foster City, Calif.). DNA templates were sequenced using a Sequenase II kit (United States Biochemical Corp., Cleveland, Ohio), and SDS-polyacrylamide gel electrophoresis (PAGE) was performed using 10-20% gradient gels (Daiichi, Tokyo, Japan). After electrophoresis, gels were fixed in 10% methanol/7% acetic acid, and soaked in autoradiography enhancer (Amplify, Amersham Arlington Heights, Ill.). After drying, autoradiography was performed overnight using X-OMAT AR2 film (Eastman Kodak, Rochester, N.Y.).
Plasmids
The vector pET-11d is available from Novagen, Inc., Madison, Wis. Plasmids were maintained and propagated in E. coli strain XL1-Blue (Stratagene, La Jolla, Calif.).
Cell Lines
K562, a human erythroleukemia-derived cell line ATCC CCL 243! known to express high levels of the human transferrin receptor at the cell surface, was cultured in RPMI 1640 medium containing 24 mM NaHCO.sub.3, 10% fetal calf serum, 2 mM glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, and 10 .mu.g/ml gentamycin. An African green monkey kidney line, Vero (ATCC CCL 81), was grown in Dulbecco's modified Eagle's medium (DMEM) supplemented as indicated above. The OKT9 hybridoma (ATCC CRL 8021), which produces a MoAb (IgG1) reactive to the human transferrin receptor, was maintained in Iscove's modified Dulbecco's medium containing 20% fetal calf serum, in addition to the supplements described above. All cell lines were cultured at 37.degree. C. in a 5% CO.sub.2 humidified atmosphere.
Construction of sFv from Hybridomas
Antibody V.sub.L and V.sub.H genes were cloned using a modification of a previously described technique (Larrick et al. Biotechniques 7:360, 1989; Orlandi et al. Proc. Natl. Acad. Sci. USA 86:3833, 1989; Chaudhary et al., 1990). Briefly, mRNA was isolated from 1.times.10.sup.8 antibody producing hybridoma cells, and approximately 3 .mu.g was reverse transcribed with M-MLV reverse transcriptase, using random hexanucleotides as primers. The resulting cDNA was screened with two sets of PCR primer pairs designed to ascertain from which Kabat gene family the heavy and light chains were derived (Kabat et al. Sequences of proteins of immunological interest. Fifth Edition. (Bethesda, Maryland: U.S. Public Health Service, 1991). Having identified the most effective primer pairs, cDNA's encoding V.sub.L and V.sub.H were spliced, separated by a region encoding a 15 amino acid peptide linker, using a previously described PCR technique known as gene splicing by overlap extension (SOE) (Johnson & Bird Methods Enzymol. 203:88, 1991). The sFv gene was then cloned into pET-11d, in frame and on the 5'-side of the PE40 gene, such that expression of the construct should generate an sFv-PE40 fusion protein approximately 70 kDa in size.
Design of Primers for PCR Amplification of V Regions Genes
The first and third complementarity determining regions (CDRs) of terminally rearranged immunoglobulin variable region genes are flanked by conserved sequences (the first framework region, FR1 on the 5' side of CDR1, and the fourth framework region, FR4, on the 3' side of CDR3).
Although murine variable region genes have been successfully cloned, regardless of family, with just two pairs of highly degenerate primers (one pair for V.sub.L and another for V.sub.H) (Gussow et al. Cold Spring Harbor Symp. Quant. Biol. 54:265, 1989; Orlandi et al., 1989; Chaudhary et al., 1990; Batra et al., 1991), the method may not be effective in cases where the number of mismatches between primers and the target sequence is extensive. With this in mind, using the Kabat database of murine V gene sequences the present invention provides a set of ten FR1-derived primers (six for V.sub.L and four for V.sub.H), such that any of the database sequences selected at random would have a maximum of three mismatches with the most homologous primer. This set of primers can be used effectively to clone V region genes from a number of MoAb secreting cell lines.
Assembly of the OKT9 sFv Gene
mRNA isolated from the hybridoma secreting the OKT9 MoAb was converted to cDNA as described previously (Larrick et al., 1989; Orlandi et al., 1989; Chaudhary et al., 1990). Despite the fact that CL-UNI is the partnering oligonucleotide in each case, a product the required size (approximately 400 bp) is not produced by V.sub.L primers IV/VI, IIa or IIb. This suggests that mismatches between these primers and the target sequence were too extensive to allow efficient amplification. A similar argument can be used to explain the failure of V.sub.H primers I and III to produce the required product. It is clear that primers V.sub.L -I/III and V.sub.H -V are most effective at amplifying the OKT9 V.sub.L and V.sub.H genes respectively. PCR amplified OKT9 V.sub.L and V.sub.H genes were spliced together using the SOE technique, as previously described (Johnson & Bird, 1991). A synthetic DNA sequence encoding a 15 amino acid linker, was inserted between the variable regions; this linker has been used very effectively in the production of functional sFv (Huston et al., 1991; Johnson & Bird, 1991), and appears to allow the variable chains to assume the optimum orientation for antigen binding. Following splicing of V region genes by the SOE procedure, the DNA fragment encoding the OKT9 sFv was electrophoresed through a 1.5% agarose gel, purified by the Geneclean technique, digested with the appropriate pair of restriction enzymes, and cloned into the pET-11d expression vector in frame and on the 5' side of the PE40 gene.
In Vitro Expression of sFv-PE40 Fusion Proteins
Plasmid templates were transcribed and translated using a rabbit reticulocyte lysate-based transcription/translation system, according to the instructions of the manufacturer, in 96-well microtiter plate format L-.sup.35 S!methionine-labeled proteins (for analysis by SDS-PAGE) and unlabeled proteins (for enzymatic analysis and bioassay), were produced in similar conditions, except that the isotope was replaced with 20 .mu.M unlabeled L-methionine in the latter case. Control lysate was produced by adding all reagents except plasmid DNA. After translation, unlabeled samples were dialysed overnight at 4.degree. C. against phosphate-buffered saline (PBS), pH 7.4 in Spectra/Por 6 MWCO (molecular weight cutoff) 50,000 tubing (Spectrum, Houston, Tex.).
Constructs incorporating the aberrant kappa transcript will contain a translation termination codon in the V.sub.L chain as previously described, and would therefore be expected to generate a translation product approximately 12 kDa in size. On the other hand, constructs which have incorporated the productive V.sub.L gene contain no such termination codon, and a full-length fusion protein (approximately 70 kDa in size) should be produced.
In vitro expression studies were used to determine the size of the protein encoded by the OKT9 sFv-PE40 gene. The constructs tested in this experiment clearly produce a protein of approximately 70 kDa, indicating that the clones do not contain the aberrant V.sub.L gene, and are devoid of frameshift mutations. Of several OKT9 sFv constructs tested, none apparently incorporated the incorrect VL gene. However, in the case of another sFv generated by this method (1B7 sFv, derived from a MoAb which binds to pertussis toxin), the majority of the clones tested produced a 12 kDa protein, and were found to contain the aberrant transcript on DNA sequencing. It should be noted that the 12 kDa fragment is frequently obscured in 10-20% gradient gels by unincorporated .sup.35 S-methionine which co-migrates with the dye front.
Determination of Protein Concentration
The enzymatic activities of fusion proteins were compared with those of known concentrations of PE in an ADP-ribosyl transferase assay, allowing molarities to be determined (Johnson et al. J. Biol. Chem. 263:1295-1399, 1988). Samples were adjusted to contain equivalent concentrations of lysate, thus maintaining an identical amount of substrate (elongation factor 2) in all cases.
Protein Synthesis Inhibition Assay for Functional sFv-PE40 Binding
Binding of the OKT9 sFv to the human transferrin receptor was qualitatively determined by assessing the ability of the OKT9 sFv-PE40 fusion protein to inhibit protein synthesis in the K562 cell line. Pseudomonas exotoxin A is a bacterial protein which is capable of inhibiting de novo protein synthesis in a variety of eukaryotic cell types. The toxin binds to the cell surface, and ultimately translocates to the cytosol where it enzymatically inactivates elongation factor 2. PE40 is a mutant form of exotoxin A which lacks a binding domain, but is enzymatically active, and capable of translocation. Fusion proteins containing PE40 and an alternative binding domain (for example, an sFv to a cell surface receptor) will inhibit protein synthesis in an appropriate cell line only if the sFv binds to a cell-surface antigen which subsequently internalizes into an acidified endosome (Chaudhary et al., 1989). The TfnR is such an antigen, so a qualitative assessment of binding may be determined by measuring the ability of the OKT9 sFv-PE40 fusion protein to inhibit protein synthesis in a cell line like K562, which expresses the TfnR. Protein synthesis inhibition assays were performed as described previously (Johnson et al., 1988). Briefly, samples were serially diluted in ice cold PBS, 0.2% BSA, and 11 .mu.l volumes were added to the appropriate well of a 96-well microtiter plate (containing 10.sup.4 cells/100 .mu.l/well in leucine-free RPMI 1640). After carefully mixing the contents of each well, the plate was incubated for the indicated time at 37.degree. C. in a 5% CO.sub.2 humidified atmosphere. Each well was then pulsed with 20 .mu.l of L-.sup.14 C(U)!leucine (0.1 .mu.Ci/20 .mu.l), incubated for 1 hour, and harvested onto glass fiber filters using a PHD cell harvester (Cambridge Technology, Cambridge, Mass.). Results are expressed as a percentage of the isotope incorporation in cells treated with appropriate concentrations of control dialyzed lysate.
The results of this assay, clearly indicate that OKT9 sFv-PE40 is capable of inhibiting protein synthesis with an IC.sub.50 (the concentration of a reagent which inhibits protein synthesis by 50%) of approximately 2.times.10.sup.-9 M. The toxicity of the fusion protein, but not of PE, was abrogated in the presence of excess OKT9 MoAb (12 .mu.g/ml), indicating that binding is specific for the TfnR. No toxicity was observed when K562 was substituted with Vero (an African Green monkey cell line which expresses the simian version of the transferrin receptor), indicating that the OKT9 sFv retains the human receptor-specific antigen binding properties of the parent antibody.
Having demonstrated binding of the OKT9 sFv to TfnR, its nucleotide sequence was determined using dideoxynucleotide chain-terminating methods, confirming extensive homology with the respective regions of immunoglobulins of known sequence.
EXAMPLE 5
Characterization of Single-Chain Antibody (sFv)-Toxin Fusion Proteins Produced in Vitro in Rabbit Reticulocyte Lysate
The present invention provides in vitro production of proteins containing a toxin domain (derived from Diphtheria toxin (DT) or PE) fused to a domain encoding a single-chain antibody directed against the human transferrin receptor (TfnR). The expression of this antigen on the cell surface is coordinately regulated with cell growth; TfnR exhibits a limited pattern of expression in normal tissue, but is widely distributed on carcinomas and sarcomas (Gatter, et al. J. Clin. Pathol. 36:539-545, 1983), and may therefore be a suitable target for immunotoxin-based therapeutic strategies (Johnson, V. G. and Youle, R. J. "Intracellular Trafficking of Proteins" Cambridge Univ. Press, Cambridge England, Steer and Hover eds., pp. 183-225; Batra et al., 1991; Johnson et al., 1988).
Proteins consisting of a fusion between an sFv directed against the TfnR and either the carboxyl-terminus 40 kDa of PE, or the DT mutant CRM 107 S(525)F! were expressed in rabbit reticulocyte lysates, and found to be specifically cytotoxic to K562, a cell line known to express TfnR. In comparison, a chimeric protein consisting of a fusion between a second DT mutant, DTM1 S(508)F, S(525)F! and the E6 sFv exhibited significantly lower cytotoxicity. Legal restrictions imposed on manipulating toxin genes in vivo previously prevented expression of potentially interesting toxin-containing fusion proteins (Federal Register 51(88)(III):16961 and Appendix F:16971); the present invention provides a novel procedure for in vitro gene construction and expression which satisfies the regulatory requirements, facilitating the first study of the potential of non-truncated DT mutants in fusion protein ITs. The present data also demonstrates that functional recombinant antibodies can be generated in vitro.
Reagents
DT and PE were purchased from List Biologicals (Campbell, Calif.). Nuclease treated, methionine-free rabbit reticulocyte lysate and recombinant ribonuclease inhibitor (rRNasin) were obtained from Promega (Madison, Wis.). Tissue culture supplies were from GIBCO (Grand Island, N.Y.) and Biofluids (Rockville, Md.). Reagents for PCR were provided by Perkin-Elmer Cetus (Norwalk, Conn.). Restriction and nucleic acid modifying enzymes were from Stratagene (La Jolla, Calif.), as was the mCAP kit used to produce capped mRNA in vitro. Geneclean and RNaid kits (for the purification of DNA and RNA respectively) were supplied by BIO 101 (La Jolla, Calif.). L-.sup.35 S!methionine, L-.sup.14 C(U)!leucine and 5'-(alpha-thio)-.sup.35 S!dATP were from New England Nuclear (Boston, Mass.). Adenylate-.sup.32 P!NAD was supplied by ICN Biomedicals (Costa Mesa, Calif.).
Oligonucleotide Synthesis
Oligonucleotides were synthesized (0.2 .mu.M scale), using cyanoethylphosphoramidites supplied by Milligen-Biosearch (Burlington, Mass.) on a dual column Cyclone Plus DNA synthesizer. Post-synthesis purification was achieved using OPC cartridges (Applied Biosystems, Foster City, Calif.).
Plasmids
pET-11d was the generous gift of Dr. F. William Studier, Brookhaven National Laboratory (Upton, N.Y.). pHB21-PE40, a derivative of pET-11d containing the gene for PE40, was kindly supplied by Dr. David FitzGerald (NIH, Bethesda, Md.). All plasmids were maintained and propagated in E. coli strain XL1-Blue (Stratagene, La Jolla, Calif.).
Cell Lines
Corynebacterium diphtheriae strain C7.sub.s (.beta.).sup.tox+ (ATCC 27012) was obtained from the ATCC (Rockville, Md.), and the strain producing the binding-deficient DT mutant CRM 103 was the generous gift of Dr. Neil Groman, University of Washington (Seattle, Wash.). Both strains were propagated in LB broth. K562 (a human erythroleukemia-derived cell line, ATCC CCL 243) was cultured in RPMI 1640 medium containing 24 mM NaHCO.sub.3, 10% fetal calf serum, 2 mM glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, and 10 .mu.g/ml gentamycin. Vero (an African green monkey kidney line, ATCC CCL 81) was grown in Dulbecco's modified Eagle's medium supplemented as described above. All eukaryotic cells were cultured at 37.degree. C. in a 5% CO.sub.2 humidified atmosphere.
Splicing Genes using PCR
Genes encoding antibody V.sub.L and V.sub.H were spliced, separated by a region encoding a 15 amino acid peptide linker, using a previously described PCR technique known as gene splicing by overlap extension (SOE) (Horton et al. Gene 77:61-68, 1989; Horton et al. Biotechniques 8:528-535, 1990). For studies requiring in vitro expression of PCR products, tox gene-derived fragments were linked to those encoding sFv using a similar method, without the use of restriction enzymes.
Construction of Plasmids Encoding Toxin-sFv Fusion Proteins
The gene encoding PE40 was obtained as an insert in pET-11d, and the sFv gene was cloned on the 5' side of this insert as indicated. To clone the gene encoding the DT binding-site mutant DTM1 S(508)F, S(525)F!, genomic DNA was isolated from the C. diphtheriae strain which produces CRM 103. DNA was extracted by a modification of the cetyltrimethylammonium bromide extraction procedure (Wilson, K. "Current Protocols in Molecular Biology" Asubel et al. eds. John Wiley & Sons New York, 2.4.1-2.4.5, 1988) and subjected to 20 cycles of PCR amplification. Primers were designed to: (i) amplify the 1605 bp region encoding CRM 103, concomitantly mutating the codon at position 525 from TCT to TTT, and (ii) incorporate restriction sites appropriate for cloning. The mutations present in CRM 107 and CRM 103 were thus combined on a single gene.
In Vitro Transcription of DNA Templates
For transcription, DNA templates required a T7 RNA polymerase promoter immediately upstream of the gene of interest (Oakley, J. L. and Coleman, J. E. Proc. Acad. Sci. U.S.A. 74:4266-4270, 1977). Such a promoter was conveniently present in pET-11d (Studier et al. Enzymol 185:60-89, 1990). In the case of PCR products, the upstream primer (a 57-mer, T7-DT) was used to introduce all of the elements necessary for in vitro transcription/translation. T7-DT includes a consensus T7 RNA polymerase promoter, together with the first seven codons of mature DT (Greenfield et al. Proc. Natl. Acad. Sci. U.S.A. 80:6853-6857, 1983) immediately preceded by an ATG translation initiation codon in the optimum Kozak context (Kozak, M. J. Biol. Chem. 266:19867-19870, 1991). m.sup.7 G(5')ppp(5')G-capped RNA was produced by transcription from linearized plasmids or PCR products using an mCAP kit, according to the manufacturer's protocol. Prior to translation, RNA was purified using an RNaid kit, recovered in nuclease free water, and analyzed by formaldehyde gel electrophoresis.
In Vitro Expression of Fusion Proteins
L-.sup.35 S!methionine-labelled proteins (for analysis by SDS-PAGE) were produced from capped RNA in methionine-free, nuclease treated rabbit reticulocyte lysate, according to the supplier's instructions. Unlabeled proteins (for bioassay), were produced in similar conditions, except that the isotope was replaced with 20 .mu.M unlabeled L-methionine. Control lysate was produced by adding all reagents except exogenous RNA. After translation, samples were dialysed overnight at 4.degree. C. against PBS, pH 7.4 in Spectra/Por 6 MWCO 50,000 tubing (Spectrum, Houston, Tex.).
Prior to transcription, plasmids were linearized at the BglII site and treated with proteinase K to destroy ribonucleases that may contaminate the sample. After phenol/chloroform extraction and ethanol precipitation, DNA was dissolved in nuclease free water to a concentration of approximately 0.2 .mu.g/.mu.l. m.sup.7 G(5')ppp(5')G-capped RNA was synthesized by T7 RNA polymerase using the conditions recommended by the manufacturer, and its integrity was confirmed by formaldehyde gel electrophoresis. Capped RNA was translated in a commercially available rabbit reticulocyte lysate, according to the instructions of the manufacturer. It is clear from the gel that the major band in each case has a molecular weight corresponding to that of the protein of interest, and that relatively large molecules (approximately 120 kDa in the case of DTM1-E6 sFv-PE40) can be synthesized in the lysate using the conditions described.
Immediately following translation, samples were extensively dialyzed overnight at 4.degree. C. against PBS, pH 7.4. The dialysis step was found to be essential, because non-dialyzed rabbit reticulocyte lysate resulted in the incorporation of significantly lower amounts of .sup.14 C-leucine upon assay by protein synthesis inhibition in all cell lines tested. After determining the concentration of the newly synthesized protein using a standard assay for measuring ADP-ribosyltransferase activity (Johnson et al., 1988), the cytotoxic activity of samples was immediately determined.
ADP-ribosyl Transferase Assay
The enzymatic activity (and therefore molarity) of fusion proteins was determined by comparison with DT or PE standard curves, as described previously (Johnson et al., 1988). Appropriate volumes of control lysate were added to each standard curve sample, in order to control for the presence of significant levels of EF-2 in reticulocyte lysate.
Other Methods
SDS-PAGE was performed as previously described (Laemmli, U. K. Nature 227:680-685, 1970), using 10-20% gradient gels (Daiichi, Tokyo, Japan). Once electrophoresis was complete, gels were fixed for 15 minutes in 10% methanol, 7% acetic acid, and then soaked for 30 minutes in autoradiography enhancer (Amplify, Amersham Arlington Heights, IL). After drying, autoradiography was performed overnight using X-OMAT AR2 film (Eastman Kodak, Rochester, N.Y.), in the absence of intensifying screens. Dideoxynucleotide chain-termination sequencing of double-stranded DNA templates was performed using a Sequenase II kit (United States Biochemical Corp., Cleveland, Ohio), according to the manufacturer's protocol.
Cytotoxicity of Toxin-sFv Fusion Proteins Expressed in Reticulocyte Lysates
The cytotoxic activity of fusion proteins was determined by their ability to inhibit protein synthesis in relevant cell lines (e.g., K562). Assays were performed as described previously (Johnson et al., 1988). Briefly, samples were serially diluted in ice cold PBS, 0.2% BSA, and 11 .mu.l volumes were added to the appropriate well of a 96-well microtiter plate (containing 10.sup.4 cells/well in leucine-free RPMI 1640). After carefully mixing the contents of each well, the plate was incubated for the indicated time at 37.degree. C. in a 5% CO.sub.2 humidified atmosphere. Each well was then pulsed with 20 .mu.l of L-.sup.14 C(U)!leucine (0.1 .mu.Ci/20 .mu.l), incubated for 1 hour, and harvested onto glass fiber filters using a PHD cell harvester (Cambridge Technology, Cambridge, Mass.). Results were expressed as a percentage of the isotope incorporation in cells treated with appropriate concentrations of control dialyzed lysate.
The results of the protein synthesis inhibition assay clearly indicate that PE40-containing fusion proteins synthesized in cell-free reticulocyte lysates are highly cytotoxic to this cell line (IC.sub.50 1.times.10.sup.-10 M). In contrast, DTM1-E6 sFv was at least ten-fold less toxic to K562 than the PE40-containing fusion protein, despite the fact that it exhibited ADP-ribosyl transferase activity indistinguishable from that of wt DT synthesized from an equivalent amount of RNA in an identical reticulocyte lysate mix. Since the decreased toxicity of DTM1-E6 sFv is clearly not due to a deficit in enzymatic activity, the binding and/or translocation process is implicated. Possible mechanisms by which the sFv-antigen interaction could be inhibited include: (i) misfolding of the sFv domain or (ii) steric interactions with other regions of the fusion protein preventing close association of sFv with the TfnR. It is of interest that a tripartite protein, DTM1-E6 sFv-PE40 was significantly cytotoxic to K562 (IC.sub.50 around 1.times.10.sup.-10 M, similar to that of PE40-E6 sFv), and the toxic effect was clearly mediated via the TfnR, since this activity was blocked by addition of excess E6 Mab. Although it is possible that the inclusion of the PE40 moiety at the carboxyl end of the tripartite molecule results in a significant conformational change in domains more proximal to the amino terminus, it seems unlikely that the sFv binding domain of DTM1-E6 is misfolded, or unavailable to interact with the TfnR. Interactions of DTM1-E6 sFv with the cell surface could be measured in a direct binding assay (Greenfield et al. Science 238:536-539, 1987), but these studies were not performed in the course of this investigation. Nevertheless, it appears likely that the lack of toxicity of the DTM1-E6 sFv fusion protein is due to a deficit in its translocation function.
The expression system developed is rapid and easy, and facilitates the manipulation of a number of samples at once. No complicated protein purification or refolding procedures are required, and the method can be used to express proteins which, due to restrictions imposed on the manipulation of toxin-encoding genes, could not be produced by more conventional methods. The technique is ideal for ascertaining the suitability of new sFv for IT development; it is theoretically possible to assemble the sFv-encoding gene (and that encoding the IT itself) by splicing of PCR products derived directly from the hybridoma, without the necessity for cloning. This would facilitate the selection of the most promising candidate molecule, prior to investing considerable effort and expense in large scale protein production and purification. Toxins and toxin-containing fusion proteins are proving to be powerful aids in our understanding of receptor mediated endocytosis and intracellular routing, and are providing valuable insight into normal cell function (reviewed in ref. 2). The method described simplifies the generation of such molecules, and facilitates their production and use in laboratories in which the application of more conventional expression methods would be impractical.
EXAMPLE 6
Cassette Mutagenesis to Produce PAHIV Mutants
Three pieces of DNA are joined together. Piece A has vector sequences and encodes the "front half" (5' end of the gene) of PA protein, B is short piece of DNA (referred to as a cassette) and encodes a small middle piece of PA protein and piece C which encodes the "back half" (3' end of the gene) of PA.
PA with alternate HIV-1 cleavage sites were created by a cassette mutagenesis procedure. Eight deoxyoligonucleotides were synthesized for construction of cassettes coding for specifically designed amino acid sequences. All four cassettes were generated by annealing two synthetic oligonucleotides (primers).
The underlined portion of each protein sequence is recognized and cleaved by the HIV-1 protease.
Primer pair 1 encodes a protein sequence which duplicates part of the cleavage site found between the membrane associated protein and the capsid protein.
Primer pair 2 encodes a protein sequence which duplicates part of the cleavage site between the capsid and the nucleocapsid protein.
Primer pair 3 encodes a protein sequence which duplicates part of the cleavage site between the protease and the p6 protein. Like the protease, p6 is a portion of the large protein produced by HIV.
Primer pair 4 encodes a protein sequence which should be cleaved by the protease. It was created by examining several protein sequences which are recognized by the HIV protease and using the common residues from each sequence. Glycine residues were added to each end to make the molecule more flexible.
The mutagenic cassettes were ligated with the BamHI/BstBI fragment from plasmid pYS5 and the PpuMI-BamI-II fragment from plasmid pYS6. Plasmids shown to have correct restriction maps were transformed into the E. coli dam- dcm- strain GM2163 (available from New England Bio-Labs, Beverly, Mass.). Unmethylated plasmid DNA was purified from each mutant and used to transform B. anthracis. For methods, see Klimpel, et al. Proc. Natl. Acad. Sci. 89:10277-10281 (1992). pYS5 and pYS6 construction are described in Singh, et al. J. Bio. Chem. 264:19103-19107 (1989).
The nucleotide and amino acid sequence of the mature PA protein after alteration with primer set 2 are shown below. Nucleotides residues 482 to 523 were replaced with cassette 2 resulting in replacement of amino acid residues 162-171 of PA with residues NTATIMMQRGNFLQ (SEQ. ID NO: 22), PAHIV#2. The altered DNA sequence and the new amino acid residues are underlined.
The above procedure was followed for PAHIV#1, 3 and 4.
EXAMPLE 7
Cleavage of Mutant PAHIV Proteins in vitro
The mutated proteins were treated with purified HIV-1 protease and evaluated for their degree of cleavage with respect to time. The purified protease was obtained from the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, Bethesda, Md. Alternatively, the protease can be purified following the method of Louis, et al., Euro. J. Biochem., 199:361 (1991).
Extended incubation (12 hours) of PA or the mutated PA proteins with the purified HIV-1 protease resulted in the appearance of two additional protein fragments that were not anticipated. These two fragments are approximately 53 kilodaltons and 30 kilodaltons in size. This may represent cleavage of PA and mutant PA proteins at a site recognized by the HIV-1 protease between PA residues y.sup.259 and p.sup.260. The residues around this cleavage site, .sup.256 VAAYPIVHV.sup.264 (SEQ. ID NO: 35), have not previously been identified as a potential HIV-1 protease cleavage site.
Incubation of RAW 264.7 cells (ATCC No. TIB 71) with lethal factor (LF) and HIV-1 protease-cleaved PAHIV#1 or PAHIV#4 caused cell death, demonstrating that the mutated PA proteins are capable of binding to LF and thus the toxic LF/PE fusion proteins. PAHIV, PAHIV#2 and PAHIV#3 have not yet been tested.
EXAMPLE 8
Evaluation of Cytotoxic Agents in Cell Cultures
The ability of the PA constructs containing the HIV-1-protease cleavage site to promote killing of HIV-1 infected cells is being evaluated in COS-1 cells (ATCC No. CRL 1650) transfected with the vector HIV-gpt. When COS cells are transfected with this plasmid vector they express all the genes for the production of HIV-1 virus particles except the envelope protein, gp160 (Page, K. A., et al., 1990. J. Virol. 64:5270-5276). Without the envelope protein the particles are not infectious. These cells express the HIV-1 proteases and properly cleave the viral protein gp55 to gp24 (Page, K. A., et al., 1990. J. Virol. 64:5270-5276). These properties make the transfected cells an excellent model system in which to evaluate the ability of protein constructs of the invention to eliminate HIV-1 infected cells from culture.
The COS-1 cells were transfected with the plasmid vector and the resulting cultures are being selected for stable transfectents. The mutated PA proteins (PAHIV#1, PAHIV#2, PAHIV#3 and PAHIV#4) are added to the culture media of growing HIV-gpt transfected COS-1 cells in the presence of the lethal factor fusion protein FP53 (Arora, N. et al. J. Biol. Chem. 267:15542 (1992)). Only cells which properly cleave the mutated PA proteins are able to bind the toxin LF fusion protein. The cultures are evaluated for protein expression (an indirect measure of viability) after 36 hours (Arora, N. and S. H. Leppla. 1992. J. Biol. Chem. 268:3334).
EXAMPLE 9
Treatment of an HIV-1 infected patient
A human patient who is infected with HIV-1 is selected for treatment. Although infected, this particular patient is asymptomatic. The patient weighs 70 kilograms. A dose of 10 micrograms per kilogram or 700 micrograms of a PAHIV in normal saline is prepared. This dosage is injected into the patient intravenously as a bolus. The dose is repeated weekly for a total of 4 to 6 dosages. The patient is evaluated regularly, such as weekly, in terms of his symptoms, physical exam and laboratory analysis according to the clinician's judgment. Tests of particular interest include the patient's complete blood count and examination for the presence of HIV infection. The treatment regimen can be repeated with or without alterations at the discretion of the clinician.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described can be used in the practice or testing of the present invention, the preferred methods and materials are now described. All publications and patent documents referenced in this application are incorporated herein by reference.
It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims.
Claims
We claim:
1. A method for targeting compounds having a desired biological activity not present on native anthrax lethal factor (LF) to a specific cell population, comprising: a) administering to the cell population a first compound comprising a first protein consisting essentially of: i) the translocation domain and the anthrax lethal factor (LF) binding domain of the native anthrax protective antigen (PA) protein, and ii) a ligand domain that specifically binds the first protein to a target on the surface of the cell population to bind the first compound to said surface; and b) administering to the resultant cell population a second compound comprising a fusion protein or conjugate consisting essentially of: i) the anthrax protective antigen (PA) binding domain of the native anthrax lethal factor (LF) protein, chemically attached to ii) a biological activity-inducing polypeptide to bind the second compound to the first compound on the surface of the cell population, internalize the second compound into the cell population, and effect the activity of the polypeptide therein.
2. A method according to claim 1, wherein the anthrax protective antigen (PA) binding domain of said second compound comprises at least the first 254 amino acid residues but less than all of the amino acid residues of the anthrax lethal factor (SEQ. ID NO: 2).
3. A method according to claim 1, wherein the ligand domain of said first compound is the ligand domain of the native anthrax protective antigen (PA) protein.
4. A method according to claim 1, wherein said second compound comprises the anthrax protective antigen (PA) binding domain of the native anthrax lethal factor (LF) protein chemically attached to a polypeptide through a peptide bond.
5. The method of claim 1, wherein the polypeptide of said second compound is a toxin.
6. The method of claim 1, wherein the polypeptide of said second compound is an enzyme.
7. The method of claim 1, wherein the ligand domain of said first compound is an antibody.
8. The method of claim 1, wherein the ligand domain of said first compound is a growth factor.
9. The method of claim 5, wherein the polypeptide of said second compound is Pseudomonas exotoxin A (PE).
10. The method of claim 5, wherein the polypeptide of said second compound is the A chain of Diptheria toxin.
11. The method of claim 5, wherein the polypeptide of said second compound is shiga toxin.
12. The method of claim 7, wherein the ligand domain of said first compound is a single chain antibody.
Non-Patent Literature (21)
- Klimpel et al, "Modified Anthrax Toxin is Cleaved . . . by HIV-1 Protease" J. Cell. Biochem, Suppl. 18B:163, abstract J515 (Jan. 1994).
- Friedlander, "Macrophages are Sensitive to Anthrax Lethal Toxin . . . " J. Biol. Chem. 261(16):7123-7126 (Jun. 1986).
- Arora et al., J. Bio. Chem., 267(22):15542-15548 (1992).
- Arora et al. Abstract: Third Internatl. Sympos. on Immunotoxins, Orlando, FL (Jun. 19-21, 1992).
- Quinn et al., J. Bio. Chem., 266(30):20124-20130 (1991).
- Oeltmann and Frankel, News, 5:2334-2337 (Jul. 1991).
- Singh et al., J. Bio. Chem., 266(23):15493-15497 (1991).
- Singh et al., J. Bio. Chem., 264(32):19103-19107 (1989).
- Leppla et al. Abstract: Fifth European Workshop on Bacterial Protein Toxins, Veldhoven (Jun. 30-Jul. 5, 1991).
- Klimpel et al. Abstract: 1992 ASM General Meeting, New Orleans, LA (May 1992).
- Novak, J.M., et al., "Functional Characterization of Protease-treated Bacillus anthracis Protective Antigen," J. of Biological Chemistry, 267(24):17186-17193 (Aug. 25, 1992).
- Ivins, B.E., et al., "Cloning and Expression of the Bacillus anthracis Protective Antigen Gene in Bacillus subtilis," Infection and Immunity, 54(2):537-542 (Nov. 1986).
- Molloy, S.S., et al, "Human Furin is a Calcium-dependent Serine Endoprotease That Recognizes the Sequence Arg-X-X-Arg and Efficiently Cleaves Anthrax Toxin Protective Antigen," J. of Biological Chemistry, 267(23):16396-16402 (Aug. 15, 1992).
- Zhang, L., et al., "Inhibition of HIV-1 RNA Production by the Diphtheria Toxin-Related IL-2 Fusion Proteins DAB.sub.486 IL-2 and DAB.sub.389 IL-2," J. of Acquired Immune Deficiency Syndromes, 5(12):1181-1187 (1992).
- O'Hare, M., et al., "Cytotoxicity of a recombinant ricin-A-chain fusion protein containing a proteolytically-cleavable spacer sequence," FEBS Lett. 273(1,2):200-204 (Oct. 1990).
- Williams, D.P., et al., "Cellular Processing of the Interleukin-2 Fusion Toxin DAB.sub.486 -IL-2 and Efficient Delivery of Diphtheria Fragment A to the Cytosol of Target Cells Requires Arg.sup.194," J. of Biological Chemistry, 265(33):20673-20677 (Nov. 25, 1990).
- Ohishi, I., et al., "Visualizations of Binding and Internalization of Two Nonlinked Protein Components of Botulinum C.sub.2 Toxin in Tissue Culture Cells," Infection and Immunity, 60(11):4648-4655 (Nov. 1992).
- Arora, Navene., et al. (1992) "Potent hybrid cytotoxins of anthrax lethal factor and the ADP-ribosylation domain of Pseudomonas exotoxin A are translocated direct to the cytosol of mammalian cells", Abstracts of the General Meeting of the American Society for Microbiology, abstract n. B-33.
- Cataldi, Angel, et al. (1992) "Regulation of pag gene expression in Bacillus anthracis: Use of a pag-lacZ transcriptional fusion", FEMS Microbiology Letters, 98(1-3):89-94.
- Klimpel, Kurt R., et al. (1992) "Anthrax toxin protective antigen is activated by a cell surface protease with the sequence specificity and catalytic properties of furin", Proceedings of the National Academy of Sciences of USA, 89(21):10277-10281.
- Arora, Navene, et al. (1993) "Residues 1-254 of anthrax toxin lethal factor are sufficient to cause cellular uptake of fused polypeptides", Journal of Biological Chemistry, 268(5):3334-3341.