440
Bloconlugate Chem. 1993, 4, 440-447
Biologically Active Interleukin 2-Ricin A Chain Fusion Proteins May Require Intracellular Proteolytic Cleavage To Exhibit a Cytotoxic Effect Jonathan P. Cook, Philip M. Savage,+ J. Michael Lord, and Lynne M. Roberts' Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK, and the Imperial Cancer Research Fund, Clinical Oncology Group, Hammersmith Hospital, London W12 OHS, UK. Received April 9, 1993'
DNA fusions encoding chimeric proteins in which human interleukin 2 (IL2) was fused to the A subunit of the plant cytotoxin ricin (RA) have been expressed in Xenopus oocytes. The constructs contained N-terminal IL2 and C-terminal RA, or N-terminal RA and C-terminal IL2. In the expressed chimeric proteins, the IL2 and RA moieties were joined by a peptide sequence containing a proteolytic cleavage site. Two proteolytically-sensitive peptide sequences were utilized; a peptide that forms the trypsinsensitive disulfide-bonded loop in diphtheria toxin (DT) or a synthetic peptide containing the factor Xa recognition site in a sequence flanked by two cysteine residues. In an in uitro cell free system the RA component was biologically active in all chimeric proteins produced since it specifically depurinated 285 ribosomal RNA. Proteolytic cleavage of the chimeras with either trypsin or factor Xa as appropriate separated the IL2 and RA moieties, but they did not remain covalently linked by a disulfide bond. Because of this, the cytotoxicity of protease-treated chimeras could not be assessed. Chimeras not pretreated with factor Xa but which contained the factor Xa target sequence were not cytotoxic to CTLL-2 cells. Rather, these molecules had a stimulatory effect that was ascribed to the IL2 moiety. In contrast, recombinant chimeric toxins containing the DT loop sequence were cytotoxic to CTLL-2 cells. Taken together the data suggest that RA-containing chimeras require intracellular proteolytic cleavage to release the RA moiety to render them cytotoxic to target cells.
INTRODUCTION Bacterial toxins such as diphtheria toxin (DT) or Pseudomonas exotoxin A (PE) possess three functional domains: a cell-binding domain responsible for binding the toxin to specific protein receptors present on the surface of target cells, a translocating domain which allows a toxic fragment to cross an intracellular membrane during endocytosis and thereby reach the cell cytosol, and a catalytically active domain responsible for the ADPribosylation of elongation factor-2 in the cytosol (1, 2). The toxic fragment containing the ADP-ribosylating domain is generated during endocytic uptake by a proteolytic cleavage step which is catalyzed by host cell protease(s). After proteolytic processing, but prior to the membrane translocation step, the toxic fragment remains covalently linked to the translocating and the cell binding domains through a single disulfide bond (3). Recombinant chimeric bacterial toxins with novel cell binding specificities have been generated by deleting the portion of the toxin gene encoding the cell binding domain and replacing it with DNA encoding an alternative cell binding peptide such as a growth factor (4,5),lymphokine (6, 7),or single chain antibody (8, 9). Such gene fusions have been expressed in Escherichia coli, purified, and, if appropriate, renatured. Provided that the new cellbinding domain and the toxin's translocating domain are linked to the ADP-ribosylating domains of the toxin via the disulfide-bonded loop containing the protease-sensitive site, the chimeric proteins are potently cytotoxic, although their target cell recognition has been altered (10). t Corresponding author: Lynne M. Roberts, Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK. Telephone: 0203 523558. Fax: 0203 523701. t Imperial Cancer Research Fund. Abstract published in Advance ACS Abstracts, October 1,
1993. 1043-1802/93/2904-0440$04.00/0
It has not proved possible to extend this approach for making single chain chimeric cytotoxins containing toxic plant proteins such as ricin. Ricin consists of a translocationally-competent toxic polypeptide, the A chain (RA) (in this case a 28s rRNA N-glycosidase (11)),and a cell binding polypeptide, the B chain, both of which initially occur together within the plant cell as a single proprotein (12). The proteolytic processing step that separates the two polypeptides occurs during toxin biogenesis within the plant (13). Processing does not involve cleavage at a serine protease-sensitive site, as is the case with the bacterial toxins, but rather involves the removal of a short linker peptide from within a disulfide-bonded loop (12). The endoprotease responsible for this processing, which requires cleavage after asparagine residues, is commonly found in plant cells but is apparently not present in mammalian cells (14). Intoxication of mammalian cells with mature plant cytotoxins, such as ricin, does not therefore require target cell proteolytic activity to release the toxic (RA) polypeptide for subsequent translocation into the cytosol. In the present paper we describe the construction of recombinant chimeric toxins in which interleukin (IL) 2, which has been successfully utilized as the cell-binding component in chimeric toxins based on both DT (5) and PE (6), has been joined to ricin A chain. However, recombinant fusion proteins containing RA are not normally cytotoxic, even when their component parts are biologically active in uitro, since proteolytic release of a translocationally-competent RA does not occur (15). To overcome this limitation we have engineered proteolytic cleavage sites into the RA-IL2 fusions to allow them to be processed either before or after cell entry in order to confer cytotoxicity. The proteolytic cleavage sites were present either in a peptide sequence that forms a naturally occurring disulfide bonded loop (from DT with a trypsin0 1993 American Chemical Society
Interleukin 2-Ricin A Chain Fusions
sensitive site) or in a synthetic peptide that contained the factor Xa (FXa) recognition site in a sequence flanked by two cysteine residues. The data presented show that while both these protease-sensitive sites were readily cleaved by the appropriate protease, the cysteine residues flanking the cleavage sites did not form a disulfide bond. In the case of the chimeric proteins containing the diphtheria toxin linker sequence, this limitation did not prevent these molecules from exhibiting a cytotoxic effect since proteolytic release of the RA component could apparently occur within target cells. The RA-IL2 fusion protein is therefore a potential cytotoxic agent for cells bearing IL2 receptors and may be developed as a useful therapeutic agent in the treatment of diseases where IL2 receptorpositive cells are involved (16). EXPERIMENTAL PROCEDURES
Materials. Restriction and DNA modification enzymes were from Gibco BRL Ltd., Paisley, UK, or Amersham International plc, Aylesbury, UK. RPMI 1640 culture medium was purchased from Gibco and [35Slmethionine (1000 Ci/mmol) was from Amersham. DNA encoding human IL2 and goat anti-IL2 antibodies were from British Biotechnology Ltd, Oxon, UK. Recombinant IL2 and soybean trypsin inhibitor were from Boehringer Mannheim UK Ltd., Lewes, UK. Trypsin, ricin, phenylmethanesulfonylfluoride and aprotinin were purchased from Sigma Chemical Company Ltd., Poole, UK, and Bio-Beads SM2 were from BioRad Laboratories Ltd., Heme1 Hempstead, UK. Aniline was from BDH Chemicals Ltd., Lutterworth, UK. Oligonucleotides encoding the factor Xa loop were produced on an Applied Biosystems DNA synthesizer, Model 380B, and those encoding the diphtheria toxin loop were provided by Greg Winter, MRC, Cambridge, UK. The oocytes were provided by H. Woodland, University of Warwick. Plasmid Construction. Oligonucleotidesencodingthe diphtheria toxin loop (DTL) (Figure 1A) were annealed and cloned into pUC18 to give pJK100. An NdeI (blunted with SI nuclease)/HindIII fragment encoding human interleukin 2 was ligated to the 3' side of DTL in pJKlOO to give pJK200. A SalIISmaI fragment encoding ricin A chain (12) was ligated with the SmaIIHindIII fragment from pJK2OO into pUC9 (SalI/HindIII) to give pJK314. An EcoRIIHindIII fragment (blunted with T4 DNA polymerase) was ligated into pSP64ABam ( 1 7) to provide an ATG start codon. The new plasmid pJK401 was used for transcription reactions utilizing the SP6 promoter and encoded ricin A chain-DTL-IL2 (RADIL). An EcoRI/ ClaI fragment encoding the full ricin signal sequence and part of RA (18) was inserted into the pJK314 similarily digested. An EcoRIIHindIII fragment encoding preRADIL (where the prefix "pre" refers to the presence of DNA encoding the ricin signal sequence) was then ligated into pGEMl and pSPJCl (below) to give pJC914G and pJC914X, respectively. Transcripts from the latter (encoding preRADIL) were used in Xenopus oocyte translocations. pSPJCl is a modified version of pSP64T (19). The multiple cloning site (MCS) and upstream Hind111 site were removed from pSP64T by blunting and religation. Annealed oligonucleotides containing a new MCS of EcoRI/NdeI/SalI/HindIII were introduced into the BglII site to give pSPJC1. The region encoding the DTL was replaced in three constructs with annealed oligonucleotides (with overhangs of KpnIIXhoI), encoding the factor Xa (FXa) recognition sequence within an unstructured loop (GlyAlaGlyAlaGly) and bounded by cysteine residues (Figure 1B). This was performed in the following plas-
Bioconjugate Chem., Vol. 4, No. 6, 1993
441
mids: pJK401, pJC914G, and pJC914X to give pJC688, pJC934G, and pJC934X respectively. pJC688 encodes RA-FXa loop-IL2 (RAXIL), and pJC934G and pJC934X encode preRAXIL. In constructing fusions carrying 1L2preceding RA DNA, a XhoI (blunted with T4 DNA po1ymerase)lHindIII fragment encoding RA was ligated to the 3' side of DTL in pJK100. The BamHI (blunted with SI nuclease)/ Hind111 fragment from this plasmid was ligated into pGEMIL2 (HpaIIHindIII)to give pJC371. This plasmid encodes IL2-DTL-RA (ILDRA). Annealed oligonucleotides (with overhangs of EcoRIIXhoI) encoding the FXa loop were ligated to the 5' side of RA. The HpaIIHindIII fragment from this plasmid was ligated into pGEMIL2 to give pJC812, which encodes IL2-FXa loop-RA (ILXRA). The EcoRIIHindIII fragments (blunted with SInuclease) from pJC371 and pJC812 were ligated to the 3'side of the full ricin signal sequence in pGEMl to give pJC971G and pJC982G, respectively. These plasmids encode preILDRA and preILXRA, respectively. EcoRIISalI fragments from these plasmids were ligated into pSPJCl to give pJC971X and pJC982X for oocyte translation experiments. Clones encoding preIL2 were constructed in a similar manner. DNA Manipulations. Standard methods for DNA manipulation were used as described (20). In vitro Transcription/Translation. Approximately 2 pg of linearized DNA was incubated in a final volume of 20 pL with 40 mM Hepes-KOH, pH 7.5, 6.5 mM magnesium acetate, 2 mM spermidine, 10mM dithiotreitol, 100 pg/mL bovine serum albumin, 0.5 mM each of ATP, CTP, and UTP, 0.1 mM GTP, 0.25 mM 7 mG(5')ppp(5')G, 20 units RNasin, and 15 units of SP6 or T7 RNA polymerase. The mixture was incubated a t 40 "C for 30 min and GTP in 20 mM Hepes-KOH, pH 7.5, was then added to a final concentration of 0.4 mM and the reaction continued for 30 min. Transcript RNA (1 pL) was translated in a wheat germ lysate system and labeled with [35Slmethionineas described (21). Samples (typically 1 2 pL) from an in uitro translation reaction were digested with trypsin or factor Xa (described below), immunoprecipitated (22)with either rabbit antiRA or goat anti-IL2 antibodies, and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE) (23). Trypsin Digestion. Samples were digested with 10 pg/mL trypsin in 50 mM Tris-HC1, pH 7.0, at 37 "C and stopped with 50 pg/mL soybean trypsin inhibitor. Factor Xa Digestion. Samples were digested with 24 pg/mL factor Xa (FXa) in 10 mM Tris-HC1, pH 7.6, 140 mM NaC1, and 1 mM EDTA for 1 h at 25 "C. Ribosome DepurinationAssay. The action of ricin A chain on rRNA is to remove an adenine base. The site of this depurination is susceptible to cleavage by aniline, releasing a 390 base RNA fragment which is indicative of RA activity. 2 pL of transcript RNA was translated in a non-nucleased reticulocyte lysate (providesthe maximum amount of substrate for RA). Proteins were solubilized by the addition of 1%(w/v) SDS and then extracted with phenolIchloroform. The RNA was then precipitated with ethanol. A 4-pg portion of RNA was incubated, in the dark, with 20 pL of 1 M aniline/acetic acid, pH 4.8 (24) at 60 "C for 2 min. The treated RNA was precipitated and resuspended in 60% (v/v) deionized formamide, 3.6 mM Tris, 3 mM sodium dihydrogen orthophosphate, and 0.2 mM EDTA. Aniline-treated and untreated RNA samples were denatured at 65 "C and analyzed on an 1.2 5% agarose-50 7% formamide gel. Xenopus Oocyte Translation. Transcripts were
442
Cook et al.
Bloconjugate Chem., Vol. 4, No. 6, 1993
translated in oocytes as described by Colman (22). A 50nL sample of transcript RNA (1mg/mL) was injected into each of approximately 30 oocytes. The oocytes were incubated overnight at 15 "C, half of them with 45 pCi [35S]methionine. The oocytes were homogenized in a buffer of 20 mM Tris-HC1, pH 7.6, 0.1 M NaCl, and 1% (v/v) Triton X-100 (25 pL per oocyte). The homogenate was treated with Bio-Beads SM2 to remove the Triton. Unlabeled samples were normalized for total protein content (measured at A ~ w ) .Two-thirds of each of the labeled homogenates was digested with trypsin or factor Xa, immunoprecipitated (22)with rabbit anti-RA and goat anti-IL2 antibodies, and analyzed by reducing/nonreducing SDS-PAGE. Cytotoxicity Assays. CTLL-2 cells (25) were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mM glutamine, 50 IU penicillin, and 50 pg/mL streptomycin plus 10 units/mL recombinant interleukin 2. Before use the cells were washed three times with fresh media and incubated for 18 h without IL2. For ricin cytotoxicity measurements cells were plated at 2 X 105 cells/mL and supplemented with 10 units/mL IL2. Doubling dilutions of whole ricin (0.05-100 ng/mL) were then added to the cells. For cytotoxicity measurements with oocyte homogenates, cells were plated at a final density of 2 X 105 cells/mL supplemented with 5 units/ mL IL2 with various dilutions of homogenates. The cells were then incubated overnight at 37 "C and 5% CO2 in air. Cells were labeled with 1 pCi [35Slmethionine for 2 h followed by harvesting and counting. Samples were analyzed in quadruplicates and expressed as a percentage of cells with no additions. RESULTS
Construction of Transcription Vectors. DNA fusions between the coding sequence of human IL2 and that of ricin A chain (RA),separated by DNA encoding a linker peptide with a protease-sensitive site, were constructed in the transcription vectors pSP64 and pGEMl as described in Experimental Procedures. The chimeric molecules encoded by these DNA fusions consisted of N-terminal RA and C-terminal IL2 or N-terminal IL2 and C-terminal RA. In each case the toxin and cytokine domains were separated by either a short amino acid sequence containing the disulfide-linked peptide loop with a trypsin-sensitive cleavage site from diphtheria toxin or a synthetic disulfidelinked peptide loop with a factor Xa-specific cleavage site (Figure 1). In a further series of constructs each fusion was preceded in frame by DNA encoding the N-terminal leader sequence (which includes the endoplasmic reticulum (ER) targeting signal sequence) from preproricin (18). Since transcripts encoding the signal sequence were to be expressed in Xenopus oocytes, they were cloned into the transcription vector pSP64T, modified as described in Experimental Procedures. In this vector, the constructs were flanked by the 5' and 3' untranslated regions of Xenopus B-globin cDNA which, upon transcription, provided stability to the RNA when injected into oocytes. DNA encoding human IL2, with or without that encoding the ricin signal sequence, was also cloned into the transcription vectors. A schematic illustration of the chimeric proteins possessing a signal sequence is shown in Figure 2. In Vitro Transcription and Translation. Transcripts were prepared for RADIL, ILDRA, RAXIL, and ILXRA in an in vitro system using the SP6 or T7 promoter and driven by SP6 or T7 RNA polymerase, respectively. Transcripts were translated in a wheat germ cell-freelysate
A 5 ' T C GAA GGT GGG TGC GCT GGT AAT AGA GTC AGA AGA TCA GTC T CCA CCC ACG CGA CCA TTA T C T CAG T C T T C T AGT CAG Glu G l y G l y C y s A l a G l y A s n A r g V a l A r g A r g Ser V a l
4 4
B 5'...
3' GGA AGC AGC CTG AGC T C T TGC GGT GGT C CCT TCG TCG GAC TCG AGA ACG CCA CCA GAG C T Gly Ser S e r Leu S e r S e r C y s G l y G l y Leu GGT GGT TGC GGT GCA GGG GCT GGC ATC GAG GGT AGG CCA CCA ACG CCA CGT CCC CGA CCG TAG CTC CCA T C C G l y G l y C y s G l y A l a G l y A l a G l y I l e Glu G l y A r g GGT GCT GGA GCA GGC TGC GGT GGA CCA CGA CCT CGT CCG ACG CCA CCT Gly A l a Gly Ala Gly Cys Gly Gly
... 3'
4
Figure 1. Oligonucleotidesencodingthe protease sensitive loops. (A) diphtheria toxin loop; (B)'factor Xa loop'. Two versions of the FXa loop were made with (i) KpnIIXhoI and (ii) HpaI/XhoI compatable ends. The arrows indicate the expected site of proteolytic cleavage.
0
Ricin A Chain
Diphtheria Toxin Loop
Ricin Signal Sequence Factor Xa Loop Interleukin 2
Figure 2. Schematic Representation of the Fusion Proteins Employed. Fusions were made with (as shown) and without the ricin signal sequence. Proteins made with an N-terminal signal sequence were prefixed 'pre'.
in the presence of [35S]methionine. The translation products were separated by SDS-PAGE and visualized by fluorography. All four transcripts gave polypeptide products of the expected molecular mass (Figure 3A, lanes 1, RADIL, 51.5 kDa; Figure 3B, Lanes 1, ILDRA, 47.5 kDa; Figure 3C, lanes 1, RAXIL, 49.5 kDa and ILXRA, 46.5 kDa). Proteolytic Cleavage and Identification of the CleavageProducts. The RADIL and ILDRA translation products were incubated with trypsin. At various times, samples were removed and further proteolysis was prevented by adding soybean trypsin inhibitor. ILDRA digestion was found to be more specific than was the case with RADIL, therefore the former was allowed to proceed for longer times. Each sample was divided into three. Material in one-third was immunoprecipitated using antibodies raised against RA, another third with antibodies against IL2, and the final third was not immunoprecipitated. Trypsin digestion produced fragments of the expected size containing either RA (-30 kDa) (Figure 3A,B, lanes RA 2-4) or IL2 (-15.5 kDa) (Figure 3A,B, lanes IL2 2-4). The RAXIL and ILXRA translation products were incubated for 60 min with factor Xa and then treated as above. Once again factor Xa digestion produced fragments of the appropriate size: RA (-30 kDa) (Figure 3C, lanes RA 2) or IL2 (-15.5 kDa) (Figure 3C, lanes IL2 2). The small amounts of fragments produced did not permit purification and N-terminal microsequence anal-
Interleukin P-Ricin A Chain Fusions
A Ab
-
1
RAD1 L
Bioconlugate Chem., VOl. 4, NO. 6, 1993 443
1I
RA IL2 1 2 3 4 1 2 3 4 M 1 2 3 4 18s
B
ILDRA
5.8s S tRNA
285
4
Figure 3. Protease sensitivity of fusion proteins. Samples from a wheat germ lysate translation were digested with protease for various times and immunoprecipitated with anti-RA (RA) or anti-IL2 (IL2) antibodies (Ab) or not immunoprecipitated (-) (see Experimental Procedures) and analyzed by SDS-PAGE: (A) RADIL digested with trypsin for 0 (lane l),4 (lane 2), 8 (lane 3), and 12 min (lane 4); (B) ILDRA digested with trypsin for 0 (lane l),8 (lane 2), 16 (lane 3), and 32 min (lane 4); and (C) RAXIL and ILXRA digested with factor Xa for 0 (lane 1)and 60 min (lane 2). Lane M, molecular mass markers increasing in sizes of 14.3, 30, 46 and 69 kDa.
ysis, so it is only assumed from sizing and antibody reactivities that cleavage occurs a t the expected sites. The Chimeric Molecules Catalytically Inactivate Ribosomes. RA kills cells by depurinating 28s rRNA a t a specific site close to the 3' end of the molecule (11). Ribosomes containing the modified 28s rRNA can no longer carry out protein synthesis. Depurination renders isolated 28s rRNA susceptible to amine-catalyzed hydrolysis of the phosphodiester bonds on either side of the modification site. This cleavage generates a small RNA fragmentof approximately 390 ribonucleotides from rabbit reticulocyte 28s rRNA. This fragment is therefore diagnostic of RA-catalyzed depurination. A variety of transcripts were translated in a rabbit reticulocyte lysate system. The products of RA transcripts (Figure 4A, lane 2; Figure 4B, lane 1) or of preRA transcripts (Figure 4A, lane 3) released the characteristic RNA fragment (arrowed) upon aniline treatment of
Figure 4. Ribosomal modification by the fusion proteins. Transcripts were translated in a rabbit reticulocytelysate system. The rRNA was extracted and briefly treated with acetic-acid aniline. A treated sample (+)was run next to an untreated sample (-) on a formamide-agarose gel. The arrow indicates the RNA fragment released by aniline treatment of modified rRNA (A) lane 1,transcript encoding yeast prepro-a factor; 2, mature ricin A chain (RA);3, preRA; 4, RADIL; 5, preRADIL; 6, RAXIL, and 7, preRAXIL, (B) lane 1,RA; 2, interleukin 2 (IL2); 3, preIL2; 4, ILDRA; 5, preILDRA; 6, ILXRA; and 7, preILXRA.
isolated rRNA; the products of preproalpha factor (Figure 4A, lane l),IL2 (Figure 4B, lane 2), and preIL2 (Figure 4B, lane 3) transcripts did not. All of the products from the RA-IL2 fusion transcripts prepared in the present study, with or without an N-terminal signal sequence, produced the diagnostic fragment (Figure 4A, lanes 4-7; Figure 4B, lanes 4-7). These data clearly show that the RA component of the fusion proteins was catalytically active, regardless of its position within the chimeric molecule. Expression in Xenopus Oocytes. Recombinant proteins were produced by microinjecting transcripts into Xenopus oocytes. In this case the transcripts, which also encoded an N-terminal signal sequence,were used so that the protein products would be directed to the lumen of the oocyte ER to facilitate folding and disulfide bond formation. The microinjected oocytes were incubatedwith [35S]methionine and the recombinant products were recovered by immunoprecipitation. Figure 5A (lane 1)shows the total [35S]-labeledproducts synthesized by oocytes which had been microinjected with preIL2 transcripts. Immunoprecipitation with antihuman IL2 antibodies recovered a product which, after reduction, ran on the gel with an apparent molecular mass of 16.5 kDa (Figure 5A, lane 2R) and was not present in homogenate from control oocytes that had been mock injected with water (data not shown). When the immu-
444
Bioconjugare Chem., Vol. 4, No. 6, 1993
A
M
2 N
1 2 R R
-4 ^-
I
Figure 5. Translation in Xenopus oocytes and protease sensitivities. Transcripts were microinjected into oocytes. After incubation the oocytes were homogenized,samples were immunoprecipitated and analyzed by SDS-PAGE under reducing (R) and nonreducing (N) conditions: (A) lane 1, total homogenate from oocytes expressing IL2; 2, equivalent sample immunoprecipitated with anti-IL2 antibodies; and M, molecular mass markers increasingin sizes of 14.3 and 30 kDa; (B)Lane 1, RADIL; 2, ILDRA; 3, ILDRA digested with trypsin; 4, RAXIL; 5, ILXRA; 6, RAXIL digested with factor Xa; 7, ILXRA digested with factor Xa; 8, RADIL digested with trypsin; 9, IL2; and M, molecular mass markers increasing in sizes of 14.3, 30 and 46 kDa.
noprecipitated product was run under nonreducing conditions, it had an increased mobility on the gel (Figure 5A, lane 2N), suggesting that the intrachain disulfide bond, essential for IL2 activity (26),was present in the recombinant IL2 molecule. Homogenates prepared from oocytes which had been microinjected with transcripts encoding preRADIL (Figure 5B, lane l),preILDRA (Figure 5B, lane 2), preRAXIL (Figure 5B, lane 4), or preILXRA (Figure 5B, lane 5) all containedrecombinantproducts of the expectedmolecular size (46-51 kDa) which were recovered by immunoprecipitation using a mixture of anti-RA and anti-IL2 antibodies. The unnicked fusions migrate in an identical fashion in the presence (shown) or absence (not shown) of reducing agent. When these products were treated with the appropriate protease, immunoprecipitated with the same antibody mixture, and run on the gels under reducing conditions, cleavage products representing both RA and IL2 components of the recombinant molecules were seen (Figure 5B, lanes 3R, 6R, 7R, and 8R). When the protease cleavage productswere electrophoresed under nonreducing conditions, the same faint fragmentswere also seen (Figure 5B, lanes 3N, 6N, 7N and 8N). What is quite clear in the nonreducing lanes is that a product of the size of the noncleaved fusion protein is absent (compare lanes 1and 8N, 2 and 3N, 4 and 6N, and 5 and 7N). This clearly indicates that the cleaved fragments did not remain covalently associated as disulfide-linked heterodimers. Apparently the two cysteine residues present in the linker peptides from either diphtheria toxin or the synthetic peptide with the factor Xa target sequence were not amenable to disulfide bond formation as the chimeric proteins folded in the lumen of the oocyte ER.
Cook et al.
Biological Activity of the Chimeric Proteins. The recombinant fusion proteins produced in Xenopus oocytes were assayed through their effect on the proliferation of the ILZdependent murine cell line CTLL-2. In order that the stimulatory effects of IL2 and the inhibitory effects of RA could be compared directly, we measured [35S]methionine incorporation into protein as an indication of proliferation. The proliferation seen with IL2 and certain of the chimeric proteins was more pronounced when measured as [3H]thymidine incorporation into DNA (data not shown). The rates of protein synthesis with the recombinant oocyte products were expressed as a percentage of that for cells treated with an equivalent volume of homogenate from control oocytes. As shown in Figure 6A, homogenates from oocytes expressing IL2 stimulated the proliferation of CTLL-2 cells, whereas whole ricin was potently cytotoxic. Homogenates from oocytes expressing either RAXIL or ILXRA were stimulatory (Figure 6B), the former to the level seen with equivalent dilutions of the IL2-containing homogenate. Presumablythe cellswere not able to proteolytically process these chimeric molecules which in this form were not cytotoxic since they were unable to reach the cytosol for the RA component to exert its effects. The stimulation seen therefore, reflects a positive effect from the IL2 component. In contrast, oocyte homogenates containing either RADIL or ILDRA inhibited cellular protein synthesis. This is ascribed to the liberation of an active RA component due to cleavage of the fusion protein by cellular protease(s). Such cleavage must generate a translocation-competent form of RA which is able to reach its ribosomal substrates in the cytosol. The inhibitory effect of RA was apparently sufficient to overcome any stimulatory effect of the IL2 component. None of the IL2-containingfusion proteins had any effect on Vero cells when compared with an oocyte homogenate control (data not shown). Since the cleavage fragments generated by proteolytic pretreatment of the recombinant chimeric molecules were not linked by a disulfide bond, their effect on CTLL-2 cells was not determined. The fragments would predictably give a stimulatory effect due to the action of the released IL2 component. DISCUSSION
We have constructed fusion proteins which contain the catalytically active subunit of the plant cytotoxin ricin (RA) and human IL2. The chimeric proteins consisted of either N-terminal RA and C-terminal IL2 or N-terminal IL2 and C-terminal RA. Previously IL2 has been fused to bacterial toxins to produce chimeric proteins in which IL2 was at the N-terminus of a fusion with PE (IL2-PE40 (6))or at the C-terminus of a fusion with DT (DABm-IL2 (5)). Both the bacterial toxin fusions were cytotoxic to cells bearing IL2 receptors (5,6). In IL2-PE40 and DAB-IL2 the toxinand IL2 moieties were separated by the protease-sensitive disulfide-bonded loop of the native cytotoxins. During endocytic uptake of bacterial toxins by mammalian cells, this loop is the site of cleavage of the toxin polypeptide by host cell protease(s). This releases the toxic domain from the cell-binding domain although they remain covalently linked via a disulfide bond. This type of target cell proteolytic processing does not occur with plant toxins such as ricin, since the cleavage step which separates the enzymatic/translocation-competent and cell binding domains of ricin occurs during toxin biosynthesis (13)and is catalyzed by a plant protease with a target-site specificity not found for mammalian proteases (14). The RA-IL2 fusions produced in the
Bloconlugate Chem., Vol. 4, No. 6, 1093 446
Interleukin 2-Rlcln A Chain Fusions
0
0.5
1
1.5
2
2.5
3
3.5
whole ricin (ng/ml)
0
0.01 0.02 0.03 0.04 0.06 0.06 0.07
Sample Dilution
Figure 6. Cytotoxicity of fusion proteins. CTLL-2 cells (2 X lo6cells/mL) were incubated with (A) doubling dilutions of whole ricin and various dilutions of oocyte homogenate containing IL2 (where 1.0 represents neat homogenate and 0.01 represents a 100-fold dilution) and (B)various dilutions of oocyte homogenate containing a fusion protein. Culture details can be found in Experimental Procedures. Figures were normalized to give homogenate only control (hom, -ve) as 100%.
present study were therefore modified to allow proteolytic separation of the RA and IL2 components either before or during uptake by mammalian cells. This was achieved by including peptide sequences containing either a trypsinsensitive cleavagesite within a naturally occurring disulfide loop (from DT (15)),or a synthetic sequence containing the factor Xa recognition site flanked by two cysteine residues potentially capable of forming a disulfide bond (Figure 1). Factor X is a serine protease of the blood clotting cascade and is normally synthesized and secreted only by liver cells. Activated factor X (Xa) is involved in the activation of prothrombin and has the target site IEGRJ. Recombinant proteins were produced by translating in vitro transcripts in a wheat germ cell-free system or, after microinjection, Xenopus oocytes. In all cases protein products of the expected size were obtained (Figures 3 and 5). Cleavage of the recombinant proteins with the appropriate protease also generated two polypeptide fragments of the expected size. The uncleaved proteins were immunoprecipitated by antibodies raised against either RA and IL2, whereas the cleavage fragments reacted with either anti-RA or anti-IL2 antibodies. The RA component of the chimeric proteins was catalytically active regardless of the position of RA within the chimeric molecules and regardless of whether such proteins possessed an uncleaved N-terminal signal sequence (Figure 4).
The protein products of transcripts microinjected into Xenopus oocyteswere directed to the ER lumen to faciliate protein folding and correct disulfide bond formation. PreIL2 transcripts produced a recombinant product which we assume contained the correct IL2 intrachain disulfide bond (between Cys58 and CyslO5). Immunoprecipitated IL2 ran with an increased mobility on gels when electrophoresed under nonreducing conditions as compared to reducing conditions (Figure 5A, lanes 2N and 2R), indicating the presence of a disulfide bond. Since this recombinant IL2 was biologically active (Figure 6A) and since the intrachain C58-C105 disulfide is essential for biologicalactivity (26),we assume that it had been correctly
formed. A similar change in gel mobility was observed when comparing reduced and nonreduced IL2 derived from the chimeric molecules after proteolytic cleavage. This suggested that the intrachain disulfide bond of IL2 was correctly formed in the chimeric proteins (Figure 5B, lanes 3,6,7, and 8). In contrast, a disulfide bond between the cysteine residues flanking the linker sequence containing the proteolytic cleavage sites was not apparently formed, since electrophoresis of proteolytically-cleaved chimeric molecules resulted in separate RA and IL2 bands under both reducing and nonreducing conditions. Previous studies in our laboratory have shown that the signal sequence of preproricin effectively delivers protein into the oocyte ER lumen (27),where the oxidizingenvironment and the presence of protein disulfide isomerase facilitate disulfide bond formation. The reason why the putative disulfide bond did not form in the chimeric molecules is unclear, particularly as the DT-derived disulfide loop allowed correct disulfide bond formation when it was included in RA-staphylococcal protein A fusions that were expressed in the periplasmic space of E . coli (15) and a factor Xa-site containing linker allowed disulfide bond formation when included in the central proricin linker (28). We assume that the folding patterns of the chimeric molecules produced in this system in the present work precluded formation of the anticipated disulfide bond. The IL2-dependent cell line CTLL-2 was stimulated by recombinant IL2 produced in oocytes and was intoxicated by whole ricin (Figure 6A). For the chimeric molecules, RAXIL and ILXRA were both stimulatory to CTLL-2 cells, presumably reflecting the biological activity of the IL2 component. Although RA is catalytically active in these chimeric molecules, we assume that the lack of an observed cytotoxic effect in CTLL-2 cells was because the RA component was not released from the chimeric molecule and as such was not able to cross an intracellular membrane in order to reach its ribosomalsubstrate. These fusions carrying the factor Xa specific site are not likely to encounter a protease with factor Xa specificity within CTLL-2 cells. Both RADIL and ILDRA were cytotoxic to CTLL-2 cells (Figure 6B) but were not cytotoxic to
440
Bloconlugate Chem., Vol. 4, No. 6, 1993
Vero cells, which lack the IL2 receptor (data not shown). We assume that the cytotoxicity resulted from intracellular cleavage at the trypsin-sensitive site. This is consistent with our findings in an earlier study in which a biologically active recombinant RA-staphylococcal protein A fusion protein was not cytotoxic to antibody-coated cells (15). Cytotoxicity was conferred by including the DT loop in the RA-protein A fusions which permitted intracellular release of the RA component (15). In this case the desired disulfide bond was formed between RA and protein A, allowing the fusions to be cleaved by trypsin treatment before cell application, an approach that was not feasible in the present study because of the failure to produce the disulfide bond. The cytotoxic effect we observed with RADIL and ILDRA was not very pronounced (a Student’s t-test indicated that, for RADIL, the dilution points 0.037,0.05, and 0.0625 were significantly different (p < 0.05) from the negative control and that the dilution point 0.037 was significantly different (p < 0.05) from the points 0.05 and 0.0625. For ILDRA, the test indicated that the dilution points 0.025, 0.037, 0.05, and 0.0625 were significantly different (p < 0.05) from the negative control and that the first three of these points were significantly different (p < 0.05) from each other and the dilution points 0.005, 0.01, and 0.015 (data not shown). This may be due to a compensatory stimulatory effect caused by the IL2 component together with the low levels of chimeric molecules produced in the oocyte system used here (approximately 1 ng per 10 oocytes). Use of a serine protease inhibitor (phenylmethanesulfonyl fluoride) and aprotinin did not improve the apparent yields upon oocyte homogenization (data not shown). Although the amounts of the toxic fusion proteins were too low to be quantified accurately, the relative amounts of each fusion used in the dilution series (shown in Figure 6) are comparable. This comparability was achieved for each fusion by assessing the amounts of [Wlmethionine-labeled immunoprecipitates from a small number of radiolabeled oocytes taken from a larger unlabeled batch. The crude homogenate volumes were then adjusted to take account of product yield variability prior to dilution and use in the cytotoxicity assessment. However, the extremely low levels of recombinant molecules produced in this expression system is clearly a serious limitation which prevented a more complete characterization of the proteolytic fragments. Because of this, we are currently reexpressing the constructs in a bacterial host. The results presented here do clearly show, however,that potentially useful chimeric toxins containing RA can be made where both the cell binding and toxic components retain biological activity, but they emphasize that the RA component must be proteolytically released within the target cell for such molecules to be cytotoxic. The intracellular compartment where proteolytic processing takes place has not been identified, but the endosome seems to be a likely site, since processing of trypsin-sensitive sequences in P E and DT has been shown to occur in this organelle. ACKNOWLEDGMENT
We wish to thank Dr. Robert Spooner for his helpful discussions and Dr. Ken Flint for performing the Student’s t-test. The factor Xa was a gift from P. Esnouf, Radcliffe
Cook et al.
Hospital, Oxford, UK. This work was supported by UK SERC grant No. GR/G/00877and a UK SERC studentship to J.P.C. LITERATURE CITED (1) Hwang, J., FitzGerald, D. J., Adhya, S., and Pastan, I. (1987) Functional domains of Pseudomonas exotoxin identified by deletion analysis of the gene expressed in E. coli. Cell 48, 129-136. (2) Olsnes, S., Kozlov,J. V., van Deurs, B., and Sandvig, K. (1991) Bacterial protein toxins acting on intracellular targets. Sem. Cell Biol. 2, 7-14. (3) Ogata, M., Chaudhary, V. K., Pastan,I., and FitzGerald, D. J. (1990) Processing of Pseudomonas exotoxin by a cellular protease results in the generation of a 37,000-Datoxinfragment that is translocated to the cytosol. J.Biol. Chem. 265,2067820685. (4) Chaudhary, V. K., FitzGerald, D. J., Adhya, S., and Pastan, I. (1987) Activity of a recombinant fusion protein between transforming growth factor type alpha and Pseudomonas exotoxin. Proc. Natl. Acad. Sci. U.S.A. 84, 4538-4542. (5) Wen, Z., Tao, X., Lakkis, F., Kiyokawa, T., and Murphy, J. R. (1991) Diphtheria toxin-related alpha-melanocyte-stimulating hormone fusion toxin. J.Biol. Chem.266,12289-12293. (6) Lorberboum-Galski, H., FitzGerald, D. J., Chaudhary, V. K., Adhya, S., and Pastan,I. (1988) Cytotoxic activity of an interleukin 2-Pseudomonas exotoxin chimeric protein produced in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 85, 1922-1926. (7) Williams, D. P., Parker, K., Bacha, P., Bishai, W., Borowski, M., Genbauffe, F., Strom, T. B., and Murphy, J. R. (1987) Diphtheria toxin receptor binding domain substitution with interleukin-2: Genetic construction and properties of a diphtheria toxin-related interleukin-2 fusion protein. Protein Eng. 1,493-498. (8) Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann, T. A., FitzGerald, D. J., and Pastan, I. (1989) A recombinant immunotoxin consisting of two antibody variable domains fused to Pseudomonas exotoxin. Nature 339, 394-391. (9) Batra, J. K., FitzGerald, D. J., Chaudhary, V. K., and Pastan, I. (1991) Single chain immunotoxins directed a t the human transferrin receptor containing Pseudomonas exotoxin A or diphtheriatoxin: Anti-TFR(Fv)-PE40 and DT388-Anti-TFR(Fv). Mol. Cell. Biol. 11, 2200-2205. (10) Pastan, I., and FitzGerald, D. J. (1991)Recombinant toxins for cancer treatment. Science 254, 1173-1177. (11) Endo, Y., and Tsurugi, K. (1987)RNAN-glycosidaseactivity of ricin A chain. Mechanism of action of the toxic lectin ricin on eukaryotic ribosomes. J. Biol. Chem. 262, 8128-8130. (12) Lamb, I., Roberts, L. M., and Lord, J. M. (1985) Nucleotide sequence of cloned cDNA coding for preproricin. Eur. J. Biochem. 148,265-270. (13) Lord, J. M. (1985)Precursors of ricin and Ricinus communis agglutinin. Eur. J . Biochem. 146, 403-409. (14) Harley, S. M., and Lord, J. M. (1985)Zn uitro endoproteolytic cleavage of castor bean lectin precursors. Plant Sci. 41,111116. (15) O’Hare, M.,Brown, A. N., Hussain, K., Gerhardt,A., Watson, G., Roberts, L. M., Vitetta, E. S., Torpe, P. E., and Lord, J. M. (1990)Recombinant Staphylococcalprotein-A ricin A chain fusion proteins containing a proteolytically cleavable spacer are cytotoxic to antibody-coated cells. FEBS Lett. 273,200204. (16) Waldmann, T. A. (1989) Multichain interleukin 2 receptor: A target for immunotherapy in lymphoma. J. Natl. Cancer Znst. 81, 914-923. (17) May, M. J., Hartley, M. R., Roberts, L. M., Kreig, P. A., Osborn, R. W., and Lord, J. M. (1989) Ribosome inactivation by ricin A chain: a sensitive method to assess the activity of wild-type and mutant polypeptides. EMBO. J. 8, 301-308. (18) Richardson, P. T., Westby, M., Roberta, L. M., Gould, J. H., Colman, A., and Lord, J. M. (1989) Recombinant proricin binds galactose but does not depurinate 28s ribosomal RNA. FEBS Lett. 255, 15-20.
Interleukin 2-Ricin A Chaln Fusions (19) Krieg,P. A., and Melton, D. A. (1984) Functional messenger RNAs are produced by SP6 in uitro transcription of cloned cDNAs. Nucleic Acid Res. 12, 7057-7070. (20) Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular cloning: a laboratory manual,2nd ed., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. (21) Anderson, C. W., Strauss, J. W., and Dudock, B. S. (1983) Preparation of a cell free protein synthesizing system from wheat germ. Methods Enzymol. 101,635-644. (22) Colman, A. (1984)Translation of eukqoticmessenger RNA in Xenopus oocytes. Transcription and translation-A practical approach (D. Hames, and S. Higgins, Eds.) pp 271302, IRL Press, Oxford. (23) Laemmli, U. K. (1970)Cleavageof structural proteins during assembly of the head bacteriophage TI. Nature 227,680-685.
Bioconlugate Chem., Voi. 4, No. 6, 1993
447
(24) d'Alessio, J. D. (1982)RNAsequencing. Gel electrophoresis of nucleis acids (D. Rickwood and B. D. Hames, Eds.) pp 173179, IRL Press, Oxford. (25) Gillis, S. and Smith, K. (1977) Long term culture of tumorspecific cytotoxic T cells. Nature 268, 154-156. (26) Robb, R. J., Kutny, R. M., Panico, M., Morris, H. R., and Chowdry, V. (1984)Aminoacidsequenceand post-translational modification of human interleukin 2. Proc. Natl. Acad. Sci. U.S.A. 81, 64864490. (27) Richardson, P. T., Gilmartin, P., Colman, A., Roberta, L. M., and Lord, J. M. (1988) Expression of functional ricin B chain in Xenopus oocytes. BiolTechnol. 6, 565-570. (28) Westby, M., Argent, R. H., Pitcher, C., Lord, J. M., and Roberts, L. M. (1992) Preparation and characterization of recombinant proricin containing an alternative proteasesensitive linker sequence. Bioconjugate Chem. 3, 375-381.
