Endocytosis, Toxins, lmmunotoxins and Viruses
Pseudomonas exotoxin: recombinant conjugates as therapeutic agents David J. FitzGerald and Ira Pastan Laboratory of Molecular Biology, Department of Health and Human Services, National Cancer Institute, Bethesda,
Maryland 20892, U.S.A.
Pseudomonas exotoxin: structure and function Pseudornonas exotoxin (PE) is a bacterial protein produced by €3. aeruginosa that is toxic for most mammalian cells [for reviews see 1, 21. The mature protein is a proenzyme and, after activation, exhibits both NAD hydrolase activity and ADP-ribosyltransferase activity. When added to mammalian cells the toxin binds to the alpha 2-macroglobulin receptor which mediates internalization to the endosoma1 compartment [ 31. In the endosome the toxin is cleaved (‘nicked’) by a cellular protease. Disulphide bond reduction then results in the release of an enzymatically active toxin fragment which translocates to the cytosol where it inhibits protein synthesis. In the cytosol; PE hydrolyses NAD and transfers the ADP-ribose moiety of NAD to elongation factor 2 (EF-2). When EF-2 is modified in this way it can no longer participate in the synthesis of new protein. The inactivation of EF-2 is lethal for the cell. Structural analysis of PE crystals by McKay and colleagues [4] revealed three prominent domains. These were termed domains I, 11, and 111: domain I was further divided into domains Ia and Ib. This study was followed first by a deletion analysis of the structural gene [5] and then by many individual studies focused on determining the location of important sequences within each domain. The results from all these studies has revealed that domain Ia (amino acids 1-252) is responsible for the toxin binding to its surface receptor; domain I1 (amino acids 253-364) is the substrate for the cellular protease and, after cleavage, has translocating activity; domain Ib (amino acids 365-399) has no known function and most of it can be deleted without loss of toxin activity; and domain I11 (amino acids 400-61 3) has the ADP-ribosylating activity and contains an endoplasmic retention sequence. A closer look at the interactions of PE with cells has revealed that a lysine at position 57 in domain I is needed for cell binding [6]. Changing this basic residue to an acidic residue such as glutamic acid or inserting a dipeptide reduces binding by 100-fold. After binding the toxin is Abbreviations used: PI.:, Pseudomonas exotoxin; EF-2, elongation factor-2; Ab, antibody.
73 I
internalized and cleaved within the endosomal compartment. Cleavage occurs in an arginine-rich loop. Specifically, the peptide bond between arginine 279 and glycine 280 is broken. The toxin is then held together by the disulphide bond that joins cystine 265 with cystine 287. While it is clear that the reduction of this bond is needed to produce an enzymatically active C-terminal fragment of 37 kDa, neither the intracellular location of the reduction step nor the mechanism of reduction are known. After reduction the 37 kDa fragment translocates to the cytosol 171. Translocation is only successful if PE has an endoplasmic retention-like sequence at its C-terminus 181. The sequence at the C-terminus of native PE is REDLK. When this is changed to KDEI, a mutant toxin is generated with slightly enhanced (3-fold) toxicity. However, if this sequence is jumbled, extended with other amino acids or removed, there is no toxicity for cells, indicating that an interaction with the endoplasmic reticulum is an important step for toxin translocation to the cytosol. In the cytosol, the C-terminal fragment ADP-ribosylates EF-2. To do this it binds NAD, hydrolyses it, and then transfers the ADPribose portion. Glutamic acid 553 has been shown to be important for NAD binding [9] (see Fig. 1 for summary of PE pathway).
Targeting PE to diseased cells Because of its potency, PE has been joined to various targeting ligands and directed to kill specific populations of cells [2, 101. In early experiments native P E was conjugated chemically to epidermal growth factor 1111 or various monoclonal antibodies [12, 131. These conjugates were active against receptor-positive or antigen-positive cells but retained some of the toxicities of native PE since no major structural changes had been imposed on the toxin. It was not until recombinant versions of the toxin that lacked the toxin’s binding domain, termed PE40, were available, that it was possible to make conjugates with high specificity for target cells and low toxicity for normal tissue [2, 141. Because PE40 lacked binding activity but retained the portions of PE responsible for translocation and enzyme activity, it was considered a suitable candidate for making toxin conjugates. Experiments in this regard involved attaching PE40
I992
Biochemical Society Transactions
Fig. I Pathway of PE within mammalian cells PE binds and is internalized from coated pit structures. When
732
PE is delivered t o the endosomal compartment processing begins. The toxin is proteolytically cleaved near Arg 279 t o produce major fragments: the N-terminal fragment is 28 kDa and the C-terminal fragment is 37 kDa. The fragments are separated by reduction of the disulphide bond that joins cystines 265 and 287. The C-terminal fragment, containing the ADP ribosylating activity then translocates t o the cytosol and inhibits protein synthesis. PE is shown with its three major domains. For simplicity, domain Ib is shown as part of domain 11. ER = endoplasmic reticulum.
c
c
a
Transit through Golgi
Trmlaation horn ER
NAD
ADP-ribylation
Of
EF-2
EF-2
either chemically or genetically to binding ligands and testing for cell killing activity. Results showed that conjugates with potent cytotoxic activities could be produced when PE40 was attached chemically to monoclonal antibodies such as HB21, which recognizes the human transferrin receptor [ 141, anti-interleukin 2 receptor (anti-Tac) which binds the p55 subunit of the interleukin 2 receptor [ 151, or B3 [ 161 which recognizes a Lewis Y carbohydrate on the surface of various carcinomas; or when it was fused genetically to cDNAs for transforming growth factor alpha [17, 181, cytokines or various single chain monoclonal antibodies [ 15, 19,201.
Construction of recombinant conjugates with PE and its derivatives The removal of domain Ia eliminates toxin binding but also produces a molecule which can be redirected simply by adding binding ligands of known specificity. As mentioned this can be done in one of two different ways. The PE40 protein can be produced in large amounts in Escherichziz coli and
Volume 20
then purified. It is then conjugated to various monoclonal antibodies. When PE40 was conjugated to the following antibodies, active immunotoxins were produced: antitransferrin receptor (HB21) [ 14, 211, anti-Tac [ 141 anti-adenocarcinoma antibody (B3) [16]. These immunotoxins were prepared by using heterobifunctional cross linking agents that formed thioether linkages. When these immunotoxins were targeted to a variety of carcinomas or leukaemic cell lines, they had potent cytotoxic activity. The removal of the DNA encoding domain Ia and the replacement of it with a specific cDNA also allowed for the construction of gene fusions. The cDNA insert then targets the toxicity of the fusion protein. This strategy has been used to construct a variety of growth factor-toxin, cytokine-toxin and antibody-toxin fusions proteins. The cDNA for the growth factor, cytokine or antibody is placed at the 5' end of the fusion. This is in contrast to the situation with the gene fusions involving diphtheria toxin where the growth factor is positioned at the 3' end [22]. Fusion proteins are expressed at high levels in E. coli. Once purified, the chimeric proteins are directed to receptors on target cells and killing is often directly related, with a small number of exceptions, to the number of receptors on the cell surface [23]. Since the growth factors retain their agonist activity, failure to achieve complete cell killing could result in undesired cell growth [24]. When cells with high levels of the receptor are being targeted, killing is often more efficient with the chimeric toxin than with native PE [18]. For example, on A431 cells transforming growth factor alpha-PE40 is 50 times more potent than native PE. Several growth factor-PE40 chimeric proteins have demonstrated cell killing activity in tumor models [25-281 and other models of human disease [29, 301. This has generated a lot of enthusiasm for this approach as a new class of therapeutic agent.
Antibody-PE conjugates and chimerics Currently antibody-PE (Ab-PE) immunotoxins are made in two different ways. PE40 is chemically attached to a native antibody molecule, or a recombinant Ab-PE40 chimeric is constructed by combining the cDNA for an Ab single chain Fv fragment with the DNA for PE40. Each type of immunotoxin has distinct properties (see Fig. 2). By analysing the results of experimentation with these immunotoxins, insights into toxin-related issues concerning cell biology, animal physiology, pathology and pharmacology can be gained. In early experiments, conjugating native PE to
Endocytosis, Toxins, lmmunotoxins and Viruses
Fig. 2 Properties of PE40 immunotoxins When PE40 is attached chemically t o a monoclonal antibody the following general properties can be expected: the size of the I : I conjugate will be 19OooO Da; the half-life after IV injection will be approximately 2-5 h; the antibody will be bivalent, and unless attachment of PE40 is close t o the combining site, there will be no loss of binding affinity; and conjugates can be made with relatively high efficiency: 20% of starting material is typical. When the P E N gene is fused with a single chain (Fv) antibody the following properties can be expected: the size will be 65000 Da; the half-life after injection will be approximately 15-20 rnin; unlike the chemical conjugates, there will be a precise attachment of antibody sequences with the toxin; there is usually a 3-5-fold decline in binding affinity due t o the rnonovalent nature of the construction: and recovery of correctly folded material is only 5- 10%.
/s-s\
Ab M, = I90 000 ‘s-c’ Random attachment Long half-life Possible damage t o combining site Fc-mediated effects PE40
Ab-Fv M, = 60 000 Precise attachment Short half-life Refolding problems No constant regions
-
~
cell targeting antibodies, invariably resulted in the creation of an immunotoxin with potent cytotoxic activity. However, the toxicity of native PE immunotoxins for the liver made their use in clinical medicine very unattractive. Because PE40 immunotoxins have more desirable properties, they have been developed as the second generation of PE immunotoxins. Single chain Fv-PE40 immunotoxins are constructed from the variable region of the heavy and light chains of an antibody molecule. The two variable regions are held together with a flexible peptide linker often composed of serine and glycine residues [ZO]. PE40 is joined to the 3’ end of the cDNA encoding either the heavy or light chain [ZO, 3 11. Recombinant immunotoxins are expressed in E. coli and purified from inclusion bodies. In contrast to chemical conjugates, where the toxin molecule is attached to the antibody molecule at residues with a primary amino group, recombinant immunotoxins are homogeneous in their primary structure. Other ways in which the chemically and genetically constructed immunotoxins differ, include the size of the final product, the lack of an
Fc segment in Fv-PE40 and the fact that the recombinant immunotoxin has only one antigen binding site (see Fig. 1). In tissue culture experiments single chain antibody Fv-PE40 immunotoxins often exhibit higher potency for target cells than do the corresponding conjugates made with whole antibody chemically linked to PE40. This result is seen despite the fact that Fv-PE40 immunotoxins have only one antigen-binding site and usually have a three-five-fold lower binding affinity for antigen compared with native antibody [20]. It is posible that the positioning of PE40 relative to the Fv is optimal for toxin processing within target cells and that this compensates for the lower binding affinity. Further experimentation and clinical evaluation is clearly needed before a full understanding can be gained both of the mechanisms of action of PE immunotoxins and their usefulness for the treatment of disease. 1. Collier, R. J. (1988) Cancer Treat. Res. 37,25-35 2. Pastan, I. & FitzCerald, D. (1989) J. Biol. Chem. 264, 15157-15160 3. Kounnas, M. Z., Morris, R. E., Thompson, M. R., FitzGerald, D. J., Strickland, D. K. & Saelinger, C. B. (1992)J. Biol. Chem. 267, 12420- 12423 4. Allured, V. S. R., Collier, R. J., Carroll, S. F. & McKay, D. B. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 1320-1 324 5. Hwang, J., FitzCerald, D. J., Adhya, S. & Pastan, I. (1987) Cell 48, 129-136 6. Jinno. Y., Chaudhary. V. K., Kondo, T.. Adhya, S., FitzGerald, D. J. & Pastan, I. (1988) J. Biol. Chem. 263,13203- 13207 7. Ogata, M., Chaudhary, V. K., Pastan, I. & FitzGerald, D. J. (1990)J. Biol. Chem. 265,20678-20685 8. Chaudhary, V. K., Jinno, Y., FitzGerald, D. J. & Pastan, I. (1990) Proc. Natl. Acad. Sci. U.S.A. 87, 308-3 12 9. Carroll, S. F. & Collier, R. J. (1987) J. Biol. Chem. 262, 8707-871 1 10. Pastan. 1. & FitzGerald, D. (1991) Science 254, 1 173- 1 177 11. FitzGerald, D. J., Padmanabhan, R., Pastan, I. & Willingham, M. C. (1983) Cell 32,607-617 12. FitzGerald, D. J., Trowbridge, 1. S., Pastan, I. & Willingham, M. C. (1983) Proc. Natl. Acad. Sci. U.S.A. 80,4134-4138 13 FitzGerald, D. J., Waldmann, T. A,, Willingham, M. C. & Pastan, I. (1984)J. Clin. Invest. 74,966-719 14. Kondo, T., FitzCerald, D. J., Chaudhary, V. K., Adhya, S. & Pastan, I. (1988) J. Biol. Chem. 263, 9470-9475 15. Kreitman, R. J., Chaudhary, V. K., Waldmann, T., Willingham, M. C., FitzGerald, D. J. & Pastan, I. (1990) Proc. Natl. Acad. Sci. U.S.A. 87,8291-8295
I992
733
Biochemical Society Transactions
7 34
16. I’ai, I,. H., Batra, J. K., FitzGerald, D. J., Willingham, M. C. & Pastan, 1. (1991) Proc. Natl. Acad. Sci. U.S.A. 88,3358-3362 17. Chaudhary, V. K., FitzGerald. L). J.. Adyha, S. & I’astan, I. (1987) I’roc. Natl. Acad. Sci. U.S.A. 84, 4538-4542 18. Siegall, C. B., Xu, Y. H., Chaudhary. V. K., Adhya, S., FitzGerald, L). J. & I’astan, I. (1989) FASEB J. 3, 2647-2652 19. Brinkman, IJ., Pai, I,. H., FitzGerald, D. J., Willingham. M. C. & Pastan, I. (1991) Proc. Natl. Acad. Sci. U S A . 88.8616-8620 20. Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann, T. A,, FitzGerald, D. J. & Pastan, I. (1989) Nature 339,394-397 21. Batra, J. K., Jinno, Y., Chaudhary, V. K.. Kondo, T., Willingham, M. C., FitzGerald, D. J. & Pastan, I. (1989) Proc. Natl. Acad. Sci. U.S.A. 86,8545-8549 22. Murphy, J. R. (1988) Cancer Treat. Res. 37, 123-40 23. Siegall, C. B., FitzCerald, L). J. & Pastan, I. (1990) J. Biol. Chem. 265, 16318-16323 24. Ogata, M., I,orberboum, G. H., FitzGerald. D. J. & Pastan, I. (1988)J. Immunol. 141,4224-4228 25. Heimbrook, L). C., Stirdivant, S. M., Ahern, J. D.,
26.
27. 28. 29.
30.
31.
Balishin, N. I,.? Patrick, D. K., Edwards. G. M., Defeo, J. D.. FitzCerald, I). J.. Pastan, I. & Oliff, A. (1990) Proc. Natl. Acad. Sci. U S A . 87, 4697-4701 Kozak, K. W., Lorberboum. G. H., Jones, I,., I’uri, K. K., Willingham, M. C., Malek, T., FitzCerald, D. J., Waldmann, T. A. & I’astan, 1. (1990)J. Immunol. 145, 2766-2771 Pai, I,. H., Gallo, M. G., FitzGerald, L). J. & Pastan, 1. (1991) Cancer Res. 51,2808-2812 Siegall, C. B., Kreitman, R. J., FitzGerald, L). J. & Pastan. I. (1991) Cancer Kes. 51,2831-2836 I,orberboum, G. H., Barrett, L. V., Kirkman, R. I,., Ogata. M., Willingham, M. C., FitzGerald, I). J. & Pastan, I. (1989) Proc. Natl. Acad. Sci. U.S.A. 86. 1008- 1012 Lorberboum-Galski, H., Lafyatis, K., Case, J. P., FitzGerald, L). J., Wilder, K. I,. & Pastan, I. (1991) Int. J. Immunopharmacol. 13. 305-3 15 Batra, J. K., Chaudhary, V. K., FitzGerald, 1). J. & Pastan. I. (1990) Biochem. Hiophys. Kes. Commun. 171.1-6
Received 23 July 1992
~~
Cell surface and intracellular functions for galactose binding in ricin cytotoxicity J. Michael Lord, Richard Wales, Carol Pitcher and Lynne M. Roberts Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, U.K.
Introduction Ricin is a heterodimeric plant protein which is potently cytotoxic to mammalian cells. Both polypeptide subunits of ricin have distinct roles in the intoxication process [ 11. The A subunit (RTA) is a ribosomal RNA N-glycosidase which, having entered the target cell cytosol, specifically removes an adenine residue from 28s rRNA (A4324in the case of rat liver 28s rRNA) [2, 31. This particular adenine residue, which is present in a highly conserved, surface-exposed loop in 28s rRNA, has a crucial role in the binding of elongation factors during translation. Ribosomes containing depurinated 28s RNA are no longer capable of protein synthesis. The ricin B subunit (RTB) is a lectin which has two galactose binding sites. Interaction between these sites and galactosides present on plasma membrane glycoproteins or glycolipids bind the holotoxin to the surface of target cells [ 11. After endocytic uptake of surface-bound toxin, the B subunit is believed to have a further function in Abbreviations used: RTA, rich A subunit; RTB, ricin B subunit; TGN, trans Golgi network; BFA, brefeldin A, EK, endoplasmic reticulum.
Volume 20
that it facilitates, either directly or indirectly, delivery of the toxic A subunit into the target cell cytosol.
Role of RTB in cytotoxicity The notion that RTB has two functional roles in the intoxication process, binding to cell surface galactose residues and facilitating RTA translocation into the cell cytosol (hereinafter referred to as the binding and translocation functions) arose from studies with ricin-containing immunotoxins. Initially RTA and RTB were separated by reducing the intrachain disulphide bond, and RTA was purified to homogeneity. The RTA was then conjugated via a disulphide bond to a monoclonal antibody raised against a cell type-specific surface antigen. In spite of binding readily to cells bearing the relevant antigen, such RTA-containing immunotoxins frequently displayed disappointingly low cytotoxicity [4]. In contrast, corresponding immunotoxins in which whole ricin was conjugated to the antibody were potently cytotoxic, even when surface binding via the RTB galactose binding sites was prevented by the addition of free galactose or lactose [S, 61. The greater cytotoxicity of whole ricin containing
Pseudomonas exotoxin: recombinant conjugates as therapeutic agents David J. FitzGerald and Ira Pastan Laboratory of Molecular Biology, Department of Health and Human Services, National Cancer Institute, Bethesda,
Maryland 20892, U.S.A.
Pseudomonas exotoxin: structure and function Pseudornonas exotoxin (PE) is a bacterial protein produced by €3. aeruginosa that is toxic for most mammalian cells [for reviews see 1, 21. The mature protein is a proenzyme and, after activation, exhibits both NAD hydrolase activity and ADP-ribosyltransferase activity. When added to mammalian cells the toxin binds to the alpha 2-macroglobulin receptor which mediates internalization to the endosoma1 compartment [ 31. In the endosome the toxin is cleaved (‘nicked’) by a cellular protease. Disulphide bond reduction then results in the release of an enzymatically active toxin fragment which translocates to the cytosol where it inhibits protein synthesis. In the cytosol; PE hydrolyses NAD and transfers the ADP-ribose moiety of NAD to elongation factor 2 (EF-2). When EF-2 is modified in this way it can no longer participate in the synthesis of new protein. The inactivation of EF-2 is lethal for the cell. Structural analysis of PE crystals by McKay and colleagues [4] revealed three prominent domains. These were termed domains I, 11, and 111: domain I was further divided into domains Ia and Ib. This study was followed first by a deletion analysis of the structural gene [5] and then by many individual studies focused on determining the location of important sequences within each domain. The results from all these studies has revealed that domain Ia (amino acids 1-252) is responsible for the toxin binding to its surface receptor; domain I1 (amino acids 253-364) is the substrate for the cellular protease and, after cleavage, has translocating activity; domain Ib (amino acids 365-399) has no known function and most of it can be deleted without loss of toxin activity; and domain I11 (amino acids 400-61 3) has the ADP-ribosylating activity and contains an endoplasmic retention sequence. A closer look at the interactions of PE with cells has revealed that a lysine at position 57 in domain I is needed for cell binding [6]. Changing this basic residue to an acidic residue such as glutamic acid or inserting a dipeptide reduces binding by 100-fold. After binding the toxin is Abbreviations used: PI.:, Pseudomonas exotoxin; EF-2, elongation factor-2; Ab, antibody.
73 I
internalized and cleaved within the endosomal compartment. Cleavage occurs in an arginine-rich loop. Specifically, the peptide bond between arginine 279 and glycine 280 is broken. The toxin is then held together by the disulphide bond that joins cystine 265 with cystine 287. While it is clear that the reduction of this bond is needed to produce an enzymatically active C-terminal fragment of 37 kDa, neither the intracellular location of the reduction step nor the mechanism of reduction are known. After reduction the 37 kDa fragment translocates to the cytosol 171. Translocation is only successful if PE has an endoplasmic retention-like sequence at its C-terminus 181. The sequence at the C-terminus of native PE is REDLK. When this is changed to KDEI, a mutant toxin is generated with slightly enhanced (3-fold) toxicity. However, if this sequence is jumbled, extended with other amino acids or removed, there is no toxicity for cells, indicating that an interaction with the endoplasmic reticulum is an important step for toxin translocation to the cytosol. In the cytosol, the C-terminal fragment ADP-ribosylates EF-2. To do this it binds NAD, hydrolyses it, and then transfers the ADPribose portion. Glutamic acid 553 has been shown to be important for NAD binding [9] (see Fig. 1 for summary of PE pathway).
Targeting PE to diseased cells Because of its potency, PE has been joined to various targeting ligands and directed to kill specific populations of cells [2, 101. In early experiments native P E was conjugated chemically to epidermal growth factor 1111 or various monoclonal antibodies [12, 131. These conjugates were active against receptor-positive or antigen-positive cells but retained some of the toxicities of native PE since no major structural changes had been imposed on the toxin. It was not until recombinant versions of the toxin that lacked the toxin’s binding domain, termed PE40, were available, that it was possible to make conjugates with high specificity for target cells and low toxicity for normal tissue [2, 141. Because PE40 lacked binding activity but retained the portions of PE responsible for translocation and enzyme activity, it was considered a suitable candidate for making toxin conjugates. Experiments in this regard involved attaching PE40
I992
Biochemical Society Transactions
Fig. I Pathway of PE within mammalian cells PE binds and is internalized from coated pit structures. When
732
PE is delivered t o the endosomal compartment processing begins. The toxin is proteolytically cleaved near Arg 279 t o produce major fragments: the N-terminal fragment is 28 kDa and the C-terminal fragment is 37 kDa. The fragments are separated by reduction of the disulphide bond that joins cystines 265 and 287. The C-terminal fragment, containing the ADP ribosylating activity then translocates t o the cytosol and inhibits protein synthesis. PE is shown with its three major domains. For simplicity, domain Ib is shown as part of domain 11. ER = endoplasmic reticulum.
c
c
a
Transit through Golgi
Trmlaation horn ER
NAD
ADP-ribylation
Of
EF-2
EF-2
either chemically or genetically to binding ligands and testing for cell killing activity. Results showed that conjugates with potent cytotoxic activities could be produced when PE40 was attached chemically to monoclonal antibodies such as HB21, which recognizes the human transferrin receptor [ 141, anti-interleukin 2 receptor (anti-Tac) which binds the p55 subunit of the interleukin 2 receptor [ 151, or B3 [ 161 which recognizes a Lewis Y carbohydrate on the surface of various carcinomas; or when it was fused genetically to cDNAs for transforming growth factor alpha [17, 181, cytokines or various single chain monoclonal antibodies [ 15, 19,201.
Construction of recombinant conjugates with PE and its derivatives The removal of domain Ia eliminates toxin binding but also produces a molecule which can be redirected simply by adding binding ligands of known specificity. As mentioned this can be done in one of two different ways. The PE40 protein can be produced in large amounts in Escherichziz coli and
Volume 20
then purified. It is then conjugated to various monoclonal antibodies. When PE40 was conjugated to the following antibodies, active immunotoxins were produced: antitransferrin receptor (HB21) [ 14, 211, anti-Tac [ 141 anti-adenocarcinoma antibody (B3) [16]. These immunotoxins were prepared by using heterobifunctional cross linking agents that formed thioether linkages. When these immunotoxins were targeted to a variety of carcinomas or leukaemic cell lines, they had potent cytotoxic activity. The removal of the DNA encoding domain Ia and the replacement of it with a specific cDNA also allowed for the construction of gene fusions. The cDNA insert then targets the toxicity of the fusion protein. This strategy has been used to construct a variety of growth factor-toxin, cytokine-toxin and antibody-toxin fusions proteins. The cDNA for the growth factor, cytokine or antibody is placed at the 5' end of the fusion. This is in contrast to the situation with the gene fusions involving diphtheria toxin where the growth factor is positioned at the 3' end [22]. Fusion proteins are expressed at high levels in E. coli. Once purified, the chimeric proteins are directed to receptors on target cells and killing is often directly related, with a small number of exceptions, to the number of receptors on the cell surface [23]. Since the growth factors retain their agonist activity, failure to achieve complete cell killing could result in undesired cell growth [24]. When cells with high levels of the receptor are being targeted, killing is often more efficient with the chimeric toxin than with native PE [18]. For example, on A431 cells transforming growth factor alpha-PE40 is 50 times more potent than native PE. Several growth factor-PE40 chimeric proteins have demonstrated cell killing activity in tumor models [25-281 and other models of human disease [29, 301. This has generated a lot of enthusiasm for this approach as a new class of therapeutic agent.
Antibody-PE conjugates and chimerics Currently antibody-PE (Ab-PE) immunotoxins are made in two different ways. PE40 is chemically attached to a native antibody molecule, or a recombinant Ab-PE40 chimeric is constructed by combining the cDNA for an Ab single chain Fv fragment with the DNA for PE40. Each type of immunotoxin has distinct properties (see Fig. 2). By analysing the results of experimentation with these immunotoxins, insights into toxin-related issues concerning cell biology, animal physiology, pathology and pharmacology can be gained. In early experiments, conjugating native PE to
Endocytosis, Toxins, lmmunotoxins and Viruses
Fig. 2 Properties of PE40 immunotoxins When PE40 is attached chemically t o a monoclonal antibody the following general properties can be expected: the size of the I : I conjugate will be 19OooO Da; the half-life after IV injection will be approximately 2-5 h; the antibody will be bivalent, and unless attachment of PE40 is close t o the combining site, there will be no loss of binding affinity; and conjugates can be made with relatively high efficiency: 20% of starting material is typical. When the P E N gene is fused with a single chain (Fv) antibody the following properties can be expected: the size will be 65000 Da; the half-life after injection will be approximately 15-20 rnin; unlike the chemical conjugates, there will be a precise attachment of antibody sequences with the toxin; there is usually a 3-5-fold decline in binding affinity due t o the rnonovalent nature of the construction: and recovery of correctly folded material is only 5- 10%.
/s-s\
Ab M, = I90 000 ‘s-c’ Random attachment Long half-life Possible damage t o combining site Fc-mediated effects PE40
Ab-Fv M, = 60 000 Precise attachment Short half-life Refolding problems No constant regions
-
~
cell targeting antibodies, invariably resulted in the creation of an immunotoxin with potent cytotoxic activity. However, the toxicity of native PE immunotoxins for the liver made their use in clinical medicine very unattractive. Because PE40 immunotoxins have more desirable properties, they have been developed as the second generation of PE immunotoxins. Single chain Fv-PE40 immunotoxins are constructed from the variable region of the heavy and light chains of an antibody molecule. The two variable regions are held together with a flexible peptide linker often composed of serine and glycine residues [ZO]. PE40 is joined to the 3’ end of the cDNA encoding either the heavy or light chain [ZO, 3 11. Recombinant immunotoxins are expressed in E. coli and purified from inclusion bodies. In contrast to chemical conjugates, where the toxin molecule is attached to the antibody molecule at residues with a primary amino group, recombinant immunotoxins are homogeneous in their primary structure. Other ways in which the chemically and genetically constructed immunotoxins differ, include the size of the final product, the lack of an
Fc segment in Fv-PE40 and the fact that the recombinant immunotoxin has only one antigen binding site (see Fig. 1). In tissue culture experiments single chain antibody Fv-PE40 immunotoxins often exhibit higher potency for target cells than do the corresponding conjugates made with whole antibody chemically linked to PE40. This result is seen despite the fact that Fv-PE40 immunotoxins have only one antigen-binding site and usually have a three-five-fold lower binding affinity for antigen compared with native antibody [20]. It is posible that the positioning of PE40 relative to the Fv is optimal for toxin processing within target cells and that this compensates for the lower binding affinity. Further experimentation and clinical evaluation is clearly needed before a full understanding can be gained both of the mechanisms of action of PE immunotoxins and their usefulness for the treatment of disease. 1. Collier, R. J. (1988) Cancer Treat. Res. 37,25-35 2. Pastan, I. & FitzCerald, D. (1989) J. Biol. Chem. 264, 15157-15160 3. Kounnas, M. Z., Morris, R. E., Thompson, M. R., FitzGerald, D. J., Strickland, D. K. & Saelinger, C. B. (1992)J. Biol. Chem. 267, 12420- 12423 4. Allured, V. S. R., Collier, R. J., Carroll, S. F. & McKay, D. B. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 1320-1 324 5. Hwang, J., FitzCerald, D. J., Adhya, S. & Pastan, I. (1987) Cell 48, 129-136 6. Jinno. Y., Chaudhary. V. K., Kondo, T.. Adhya, S., FitzGerald, D. J. & Pastan, I. (1988) J. Biol. Chem. 263,13203- 13207 7. Ogata, M., Chaudhary, V. K., Pastan, I. & FitzGerald, D. J. (1990)J. Biol. Chem. 265,20678-20685 8. Chaudhary, V. K., Jinno, Y., FitzGerald, D. J. & Pastan, I. (1990) Proc. Natl. Acad. Sci. U.S.A. 87, 308-3 12 9. Carroll, S. F. & Collier, R. J. (1987) J. Biol. Chem. 262, 8707-871 1 10. Pastan. 1. & FitzGerald, D. (1991) Science 254, 1 173- 1 177 11. FitzGerald, D. J., Padmanabhan, R., Pastan, I. & Willingham, M. C. (1983) Cell 32,607-617 12. FitzGerald, D. J., Trowbridge, 1. S., Pastan, I. & Willingham, M. C. (1983) Proc. Natl. Acad. Sci. U.S.A. 80,4134-4138 13 FitzGerald, D. J., Waldmann, T. A,, Willingham, M. C. & Pastan, I. (1984)J. Clin. Invest. 74,966-719 14. Kondo, T., FitzCerald, D. J., Chaudhary, V. K., Adhya, S. & Pastan, I. (1988) J. Biol. Chem. 263, 9470-9475 15. Kreitman, R. J., Chaudhary, V. K., Waldmann, T., Willingham, M. C., FitzGerald, D. J. & Pastan, I. (1990) Proc. Natl. Acad. Sci. U.S.A. 87,8291-8295
I992
733
Biochemical Society Transactions
7 34
16. I’ai, I,. H., Batra, J. K., FitzGerald, D. J., Willingham, M. C. & Pastan, 1. (1991) Proc. Natl. Acad. Sci. U.S.A. 88,3358-3362 17. Chaudhary, V. K., FitzGerald. L). J.. Adyha, S. & I’astan, I. (1987) I’roc. Natl. Acad. Sci. U.S.A. 84, 4538-4542 18. Siegall, C. B., Xu, Y. H., Chaudhary. V. K., Adhya, S., FitzGerald, L). J. & I’astan, I. (1989) FASEB J. 3, 2647-2652 19. Brinkman, IJ., Pai, I,. H., FitzGerald, D. J., Willingham. M. C. & Pastan, I. (1991) Proc. Natl. Acad. Sci. U S A . 88.8616-8620 20. Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann, T. A,, FitzGerald, D. J. & Pastan, I. (1989) Nature 339,394-397 21. Batra, J. K., Jinno, Y., Chaudhary, V. K.. Kondo, T., Willingham, M. C., FitzGerald, D. J. & Pastan, I. (1989) Proc. Natl. Acad. Sci. U.S.A. 86,8545-8549 22. Murphy, J. R. (1988) Cancer Treat. Res. 37, 123-40 23. Siegall, C. B., FitzCerald, L). J. & Pastan, I. (1990) J. Biol. Chem. 265, 16318-16323 24. Ogata, M., I,orberboum, G. H., FitzGerald. D. J. & Pastan, I. (1988)J. Immunol. 141,4224-4228 25. Heimbrook, L). C., Stirdivant, S. M., Ahern, J. D.,
26.
27. 28. 29.
30.
31.
Balishin, N. I,.? Patrick, D. K., Edwards. G. M., Defeo, J. D.. FitzCerald, I). J.. Pastan, I. & Oliff, A. (1990) Proc. Natl. Acad. Sci. U S A . 87, 4697-4701 Kozak, K. W., Lorberboum. G. H., Jones, I,., I’uri, K. K., Willingham, M. C., Malek, T., FitzCerald, D. J., Waldmann, T. A. & I’astan, 1. (1990)J. Immunol. 145, 2766-2771 Pai, I,. H., Gallo, M. G., FitzGerald, L). J. & Pastan, 1. (1991) Cancer Res. 51,2808-2812 Siegall, C. B., Kreitman, R. J., FitzGerald, L). J. & Pastan. I. (1991) Cancer Kes. 51,2831-2836 I,orberboum, G. H., Barrett, L. V., Kirkman, R. I,., Ogata. M., Willingham, M. C., FitzGerald, I). J. & Pastan, I. (1989) Proc. Natl. Acad. Sci. U.S.A. 86. 1008- 1012 Lorberboum-Galski, H., Lafyatis, K., Case, J. P., FitzGerald, L). J., Wilder, K. I,. & Pastan, I. (1991) Int. J. Immunopharmacol. 13. 305-3 15 Batra, J. K., Chaudhary, V. K., FitzGerald, 1). J. & Pastan. I. (1990) Biochem. Hiophys. Kes. Commun. 171.1-6
Received 23 July 1992
~~
Cell surface and intracellular functions for galactose binding in ricin cytotoxicity J. Michael Lord, Richard Wales, Carol Pitcher and Lynne M. Roberts Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, U.K.
Introduction Ricin is a heterodimeric plant protein which is potently cytotoxic to mammalian cells. Both polypeptide subunits of ricin have distinct roles in the intoxication process [ 11. The A subunit (RTA) is a ribosomal RNA N-glycosidase which, having entered the target cell cytosol, specifically removes an adenine residue from 28s rRNA (A4324in the case of rat liver 28s rRNA) [2, 31. This particular adenine residue, which is present in a highly conserved, surface-exposed loop in 28s rRNA, has a crucial role in the binding of elongation factors during translation. Ribosomes containing depurinated 28s RNA are no longer capable of protein synthesis. The ricin B subunit (RTB) is a lectin which has two galactose binding sites. Interaction between these sites and galactosides present on plasma membrane glycoproteins or glycolipids bind the holotoxin to the surface of target cells [ 11. After endocytic uptake of surface-bound toxin, the B subunit is believed to have a further function in Abbreviations used: RTA, rich A subunit; RTB, ricin B subunit; TGN, trans Golgi network; BFA, brefeldin A, EK, endoplasmic reticulum.
Volume 20
that it facilitates, either directly or indirectly, delivery of the toxic A subunit into the target cell cytosol.
Role of RTB in cytotoxicity The notion that RTB has two functional roles in the intoxication process, binding to cell surface galactose residues and facilitating RTA translocation into the cell cytosol (hereinafter referred to as the binding and translocation functions) arose from studies with ricin-containing immunotoxins. Initially RTA and RTB were separated by reducing the intrachain disulphide bond, and RTA was purified to homogeneity. The RTA was then conjugated via a disulphide bond to a monoclonal antibody raised against a cell type-specific surface antigen. In spite of binding readily to cells bearing the relevant antigen, such RTA-containing immunotoxins frequently displayed disappointingly low cytotoxicity [4]. In contrast, corresponding immunotoxins in which whole ricin was conjugated to the antibody were potently cytotoxic, even when surface binding via the RTB galactose binding sites was prevented by the addition of free galactose or lactose [S, 61. The greater cytotoxicity of whole ricin containing
