A Chimeric Fusion Protein Containing Transforming Growth Factor-Α

Home , Fusion gene, Fusion protein

Gene Therapy (1998) 5, 521–530  1998 Stockton Press All rights reserved 0969-7128/98 $12.00 http://www.stockton-press.co.uk/gt A chimeric fusion protein containing transforming growth factor-␣ mediates gene transfer via binding to the EGF receptor

J Fominaya1,2, C Uherek1 and W Wels1 1Institute for Experimental Cancer Research, Tumor Biology Center, Breisacher Strasse 117, D-79106 Freiburg, Germany

Fusion proteins engineered to incorporate distinct functions binding to EGF receptors. Complexes of the chimeric pro- which co-operate in mediating the cell-type specific uptake tein and plasmid DNA carrying a luciferase reporter gene, and intracellular delivery of DNA present an attractive after condensation with poly-L-lysine resulted in an up to approach for the development of self-assembling vector 150-fold increase in reporter gene expression in EGF systems for targeted gene transfer. Here we have chosen receptor expressing cells in comparison to poly-L-lysine– the EGF receptor overexpressed in many human tumors DNA complexes alone. While in COS-1 cells no of epithelial origin as a target for a novel modular fusion additional endosome escape activity was required, in protein. We have fused a cDNA fragment of the human A431 cells gene delivery was dependent on the simul- EGF receptor ligand TGF-␣ to sequences encoding the taneous presence of the endosome destabilizing translocation domain of Pseudomonas exotoxin A as an reagent chloroquine indicating that cell-type specific fac- endosome escape activity, and the DNA-binding domain of tors such as different intracellular routing of protein–DNA the yeast GAL4 transcription factor. Upon bacterial complexes greatly influence transfection efficiency. expression, this TEG fusion protein displayed specific

Keywords: gene therapy; non-viral gene delivery; fusion protein; TGF-␣; EGF receptor; Pseudomonas exotoxin A; GAL4

Introduction distinct activity required for DNA-binding, cell recognition and intracellular delivery. Such fusion proteins in The development of vectors for target-cell specific gene contrast to chemical conjugates can be produced in suit- delivery is a major goal of gene therapeutic strategies. able expression systems in their final form and the Thereby nonviral gene delivery vectors are gaining resulting products are generally homogeneous in their 1,2 increasing interest. While they are still relatively inef- composition. We have recently described such a chimeric ficient, they might present a suitable alternative to viral multidomain fusion protein which facilitates target-cell vectors since they overcome major disadvantages of specific gene delivery and is based on the molecular viruses such as the lack of cell type specificity with organization of the bacterial Pseudomonas exotoxin A regard to their infectivity, the restriction in the size of (ETA).6 incorporated DNA, and safety considerations due to the The 66 kDa exotoxin A consists of three functional presence of viral genomic sequences. Progress has been domains defined by crystal structure and deletion analy- made in the understanding of how individual activities sis.8,9 Upon cell binding via the N-terminal cell recog- of viruses can be mimicked and methodologies have been nition domain (Ia) and receptor-mediated internalization, developed which allow the combination of different func- ETA is cleaved by a cellular protease within the internal tions required for gene transfer into an artificial complex. domain II (translocation domain) resulting in an N-ter- Molecular conjugates which employ antibodies or minal 28 kDa and a C-terminal 37 kDa fragment. After other ligands chemically coupled to polycations to reduction of a disulfide bond, the C-terminal fragment deliver DNA into cells via receptor-mediated endocytosis containing the enzymatic domain III translocates to the constitute an important group of such target-cell specific, cytoplasm where it ADP-ribosylates and inactivates 3,4 nonviral vectors. Another attractive approach for the eukaryotic elongation factor 2.10 In the 5EG fusion pro- design of modular self-assembling systems for gene tein,6 the natural cell binding domain of ETA was delivery is based on fusion proteins engineered to incor- replaced by a tumor-specific single chain antibody porate several co-operating functions in a single polypep- domain. Likewise, the C-terminal enzymatic domain of 5–7 tide chain. Each of these domains can account for a the toxin was exchanged with the DNA-binding domain of the yeast GAL4 transcription factor for the interaction with plasmid DNA. The internal translocation domain of Correspondence: W Wels ETA was retained as an endosome escape function. The 2Present address: Department of Immunology and Oncology, Centro ´ ´ Nacional de Biotecnologıa/CSIC, Universidad Autonoma de Madrid, resulting chimeric molecule was non-toxic, exhibited an Campus Cantoblanco, Madrid E-28049, Spain altered target–cell tropism, and functioned as a cell- Received 5 September 1997; accepted 28 November 1997 specific DNA carrier. Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 522 Due to this modular organization, similar carrier proteins with modified activities, eg different target cell specificity, might be obtained by replacing individual protein domains. This could assist in the development of gene transfer strategies optimized for distinct target cell populations. In order to explore the flexibility of this concept, we have constructed and characterized a multidomain fusion protein which employs the natural EGF receptor ligand transforming growth factor (TGF)-␣ as a cell targeting domain. EGF receptor overexpression has been observed in a high percentage of human carcinomas including glioblastoma, cancer of the lung, breast, head and neck and bladder, and correlates with an unfavorable prognosis for the patients.11 Therefore the EGF receptor constitutes an attractive target for directed tumor therapy which includes gene therapeutic strategies.12–14 We have fused a human TGF-␣ cDNA fragment to sequences encoding the ETA translocation domain and the GAL4 DNA-binding domain. Upon bacterial expression and purification, the resulting chimeric protein termed TEG displayed specific binding to EGF receptor expressing cells. Complexes of the fusion protein with plasmid DNA carrying a luciferase reporter gene and a GAL4 specific recognition sequence, after condensation with poly-l-lysine were able to transfect EGF receptor expressing cells in vitro. Thereby transfection efficiency and dependence on the endosomolytic reagent chloro- Figure 1 (a) Schematic representation of the TEG fusion gene in the E. quine varied in different cell lines demonstrating the coli expression plasmid pWF47-TEG. The inducible tac promoter controls importance of cell type-specific characteristics for suc- the expression of the chimeric TEG protein which consists of an N-ter- cessful gene delivery. minal E. coli ompA signal peptide (SP), the synthetic FLAG epitope (F), amino acids 1 to 50 of human TGF-␣, a polyhistidine tag (H), amino acids 252 to 366 of Pseudomonas exotoxin A (translocation domain) Results amino acids 2 to 147 of the yeast GAL4 protein (DNA-binding domain) and a C-terminal KDEL endoplasmic reticulum retention signal. (b) SDS- Construction and bacterial expression of the TEG DNA PAGE analysis of the TEG protein purified from bacterial lysate. Lane 1, carrier protein bacterial lysate cleared by ultracentrifugation; lane 2, flow through from 2+ DNA fragments encoding amino acids 1 to 50 of human aNi -saturated chelating Sepharose column; lane 3, wash with 50 mm TGF-␣, domain II (translocation domain) of Pseudomonas imidazole; lane 4, eluate with 250 mm imidazole; lane 5, immunoblot analysis of purified TEG protein with a MAb specific for the GAL4 DNA exotoxin A, and the DNA binding domain of the yeast binding domain; M, molecular weight standards. transcriptional activator GAL4 were derived from previously described plasmids by restriction enzyme digestion and assembled into a single open reading frame in Binding properties of the TEG fusion protein the bacterial expression vector pFLAG-1. The resulting To investigate the functionality of the TGF-␣ domain, the plasmid pWF47-TEG encodes under the control of the binding of purified TEG protein to the recombinant EGF IPTG inducible tac promoter the chimeric fusion protein receptor was analyzed by ELISA. Purified TEG at concen- termed TEG which is schematically shown in Figure 1a. trations ranging from 0.05 to 333 nm was added to the It consists of an N-terminal E. coli ompA signal peptide, wells of 96-well plates coated with recombinant extra- a synthetic FLAG epitope, human TGF-␣, a cluster of six cellular domain of the EGF receptor, the plates were incu- histidine residues facilitating the purification of the bated at room temperature for 1 h and specifically bound recombinant protein via Ni2+ affinity chromatography, protein was determined. The results are shown in Figure amino acids 252 to 366 of Pseudomonas exotoxin A, and 2a. TEG bound to the extracellular portion of the EGF amino acids 2 to 147 of the yeast GAL4 protein. At the receptor in a saturable fashion. This interaction with the C-terminus, the fusion protein carries a KDEL signal EGF receptor was dependent on the TGF-␣ domain of which is functionally equivalent to the C-terminal TEG since the presence of a nine- to 40-fold molar excess sequence of exotoxin A required for intracellular routing of TGF␣-ETA, a recombinant toxin containing TGF-␣,16 and full activity of the wild-type toxin.15 inhibited TEG binding almost completely (Figure 2b). The fusion protein was expressed in E. coli strain The interaction of TEG with EGF receptor was further CC118 and purified under denaturing conditions from confirmed by FACS analysis. A431 human epidermoid bacterial lysates via binding to Ni2+-saturated chelating carcinoma cells which highly overexpress EGF receptor sepharose and elution with 250 mm imidazole. Fractions were incubated with TEG, and specifically bound fusion containing the recombinant fusion protein were pooled, protein was detected with MAb M2 recognizing a syn- imidazole and denaturant were removed by dialysis and thetic FLAG epitope included in the recombinant protein the protein was concentrated by ultrafiltration. SDS- and FITC-labeled goat anti-mouse IgG. The results are PAGE analysis of the 38 kDa TEG fusion protein revealed shown in Figure 2c. Significant binding of TEG protein a purity of approximately 90% after a single round of to A431 cells was detected. purification (Figure 1b). Upon cell binding, internalization of the ligand–recep- Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 523

Figure 2 Binding of the TEG fusion protein to EGF receptor. (a) Recombinant protein comprising the extracellular domain of human EGF receptor was immobilized on 96-well microtiter plates and incubated with increasing concentrations of purified TEG protein as indicated. The amount of specifically bound protein was measured after incubation with a MAb specific for the GAL4 DNA binding domain followed by alkaline phosphatase coupled goat anti-mouse IgG and conversion of the phosphatase substrate p-nitrophenylphosphate as the absorbance at 405 nm. (b) The specificity of binding was confirmed by competing the binding of TEG to immobilized EGF receptor protein with an excess of the TGF-␣-containing exotoxin A fusion protein TGF␣-ETA16 as indicated. (c) The binding of TEG to the surface of EGF receptor overexpressing A431 human epidermoid carcinoma cells was detected by FACS analysis with MAb M2 directed to the synthetic FLAG epitope of the fusion protein and FITC-labeled goat anti-mouse IgG. tor complex is required for the DNA–carrier protein to shown in Figure 3. Treatment of cells with EGF (lanes 2 be effective in gene delivery. Binding of growth factors and 6) or TEG (lanes 3 and 7) led to a strong increase in like EGF and TGF-␣ to the EGF receptor triggers receptor the phosphotyrosine content of a protein corresponding dimerization and autophosphorylation of specific tyro- in size with the 170 kDa EGF receptor, which was con- sine residues within the receptor’s intracellular domain. firmed by reprobing the filter with an anti-EGF receptor This initiates the signal transduction cascade and also serum (Figure 3, lower panel). This TEG-mediated acti- results in the recruitment of the EGF receptor to clathrin- vation of the receptor was specific since it was blocked coated pits which is followed by internalization of the by the addition of the EGF receptor specific MAb 14E1 ligand–receptor complex along the endocytic pathway.17 (lanes 4 and 8). PBS and the competing antibody alone To determine whether binding of TEG to the EGF recep- had no effect on the activation status of the receptor tor induces a response similar to that initiated by natural (lanes 1, 5 and 9). These results show that TEG like the ligands, the ability of TEG to activate EGF receptor was natural EGF receptor ligand initiates receptor activation tested. COS-1 and A431 cells were treated for 10 min at which is required for enhanced internalization of the 37°C with the fusion protein in the presence or absence ligand–receptor complex. of the antagonistic EGF-receptor specific MAb 14E116 as a competitor. Control cells were treated with PBS, MAb The TEG carrier protein facilitates gene transfer 14E1 alone, or 100 ng/ml EGF. Equal amounts of cell TEG carrier protein-mediated gene transfer was analyzed lysates were assayed for their phosphotyrosine content using the plasmid pSV2G4LUC,6 which encodes a lucifer- by SDS-PAGE and subsequent immunoblotting with a ase reporter gene under the control of the SV40 early pro- specific anti-phosphotyrosine antibody. The results are moter. The plasmid contains two consecutive GAL4- Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 524

Figure 3 Activation of the EGF receptor by the TEG fusion protein. COS-1 (lanes 1–4) and A431 cells (lanes 5–9) were grown in low serum for 8 h and then treated with 3 ␮g/ml TEG in the absence (lanes 3 and 7) or the presence (lanes 4 and 8) of the antagonistic MAb 14E1. Control cells were treated with PBS (lanes 1 and 5), 100 ng/ml EGF (lanes 2 and 6) or 14E1 antibody (lane 9). Equal amounts of cell lysates were analyzed by SDS- PAGE and immunoblotting with an anti-phosphotyrosine MAb, followed by incubation with an anti-mouse horseradish peroxidase labeled antibody and chemiluminescent detection (upper panel). The amount of EGF receptor loaded in each lane was analyzed by reincubation of the ﬁlter with 12E EGF receptor speciﬁc antiserum (lower panel). The position of the 170 kDa EGF receptor is indicated (EGFR). M, molecular weight standards.

specific recognition sequences in the 3Ј-untranslated Figure 4, significant luciferase activity could already be region of the expression cassette which allows specific detected in cells treated with TEG-polyplex for 1 h. binding of fusion proteins containing the GAL4 DNA- Thereafter the expression of the reporter gene increased binding domain. TEG fusion protein and circular with incubation time in a saturable manner indicating pSV2G4LUC plasmid-DNA were mixed and incubated that effective delivery of the plasmid–DNA was for 15 min at room temperature to allow the formation of achieved. protein–DNA complexes. Subsequently an electroneutral In different experiments, we observed a relatively high complex was achieved by adding to the preincubated variation in the efficiency of gene transfer ranging from × 6 × 8 mixture a 50-fold molar excess of poly-l-lysine pL236 with 5 10 to 1 10 relative luciferase units per milligram respect to the DNA concentration.6 protein when different batches of TEG carrier protein Transfection mediated by the TEG-DNA–poly-l-lysine were used. Thereby, in contrast to the previously complex (TEG–polyplex18) was analyzed using COS-1 described ErbB2-specific carrier protein 5EG6 usually SV40-transformed monkey kidney cells in standard relatively high concentrations of TEG (50–160 nm final growth medium. These cells express approximately 1– concentration) were required for successful transfection. 2 × 105 EGF receptors per cell (W Wels, unpublished data) This suggests that similar to previously described recom- which are activated upon binding of TEG as shown in binant toxins containing TGF-␣19 a major part of the pur- Figure 3. COS-1 cells were incubated with TEG- ified carrier protein was inactive. Therefore an additional pSV2G4LUC polyplex for different time intervals, the unfolding/refolding step was included in the purification growth medium was replaced, and the activity of the protocol which increased the yield of functionally active reporter gene was analyzed 48 h after initial sample TEG fusion protein and resulted in consistently high addition. Cells incubated with a similar polyplex lacking levels of reporter gene expression using lower concen- the TEG fusion protein served as a control. As shown in trations (0.8–4 nm final concentration) of carrier protein. Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 525

Figure 4 Time course of TEG-mediated gene transfer. COS-1 cells were seeded in 12-well tissue culture plates at a density of 4 × 104 cells per well and grown overnight. Transfection complexes containing the TEG fusion protein, pSV2G4LUC luciferase reporter plasmid and poly-l-lysine (TEG-pSV2G4LUC polyplex) were prepared as described in Materials and methods and added to the cells in normal growth medium. The cells were incubated with TEG-pSV2G4LUC polyplex (TEG) at a ﬁnal concentration of partially refolded TEG of 160 nm for 10 s, or 0.5, 1, 2, 4 or 21 h as indicated. Control cells were treated with pSV2G4LUC polyplex lacking the fusion protein (pL). Cells were harvested and reporter gene expression was analyzed 48 h after the addition of polyplex. Luciferase activity is expressed in relative light units per milligram of total protein.

TEG-mediated gene transfer is dependent on specific binding to DNA and to the EGF receptor Complexes of TEG and pSV2G4LUC plasmid–DNA were prepared as described above and, after condensation with poly-l-lysine were added to COS-1 cells in normal growth medium for 16 h. The medium was replaced and luciferase activity was analyzed 36 h later. As shown in Figure 5 (a) Efficient gene transfer is dependent on the presence of TEG Figure 5a, under these conditions cells treated with TEG- carrier protein. COS-1 cells were treated with TEG-pSV2G4LUC polyplex polyplex displayed up to 150-fold higher reporter gene (TEG) at a final concentration of partially refolded TEG of 50 nm for 16 h as described in the legend to Figure 4. Control cells were incubated with activity than control cells treated with polyplex lacking pSV2G4LUC polyplex lacking the fusion protein (pL). Cells were har- the carrier protein. To investigate the dependence of vested and reporter gene expression was analyzed 36 h later. Luciferase TEG-mediated gene transfer on the specific interaction of activity is expressed in relative light units per milligram of total protein. the GAL4 DNA-binding domain with its cognate DNA (b) TEG-mediated gene transfer is dependent on specific interaction of recognition sequence, COS-1 cells were treated with poly- the GAL4 DNA-binding domain. COS-1 cells were treated with polyplex plex containing TEG and either pSV2G4LUC or the simi- containing partially refolded TEG at a final concentration of 80 nm and either pSV2G4LUC or the similar pSV2LUC plasmid which lacks a lar pSV2LUC plasmid which lacks GAL4-specific DNA GAL4-specific DNA recognition sequence. Control cells were incubated recognition sequences. As shown in Figure 5b in the with pSV2G4LUC or pSV2LUC polyplex lacking the fusion protein. Cells absence of a GAL4-specific recognition sequence the were harvested and reporter gene expression was analyzed as described efficiency of TEG-mediated gene transfer is drastically above. (c) TEG-mediated gene transfer is dependent on EGF receptor bind- reduced demonstrating the requirement of specific pro- ing. COS-1 cells were treated with TEG-pSV2G4LUC polyplex at a final tein–DNA interaction for carrier protein-mediated gene concentration of refolded TEG of 0.8 nm in the absence or presence of a 30-fold molar excess of the EGF-receptor specific antibody 14E1 or a 10- transfer. To evaluate the specificity of TEG-mediated fold molar excess of EGF as indicated. Cells were harvested and reporter gene delivery further, the effect of specific competitors gene expression was analyzed as described above. Relative luciferase activi- on luciferase expression were analyzed. COS-1 cells were ties in comparison to cells treated in the absence of competitor are shown. treated with TEG-polyplex either in the absence or presence of a 30-fold molar excess of the EGF-receptor spe- confirm that TEG-mediated gene transfer is strictly cific antibody 14E1 (24 nm final concentration), or a 10- dependent on the accessibility of EGF receptor on the sur- fold molar excess of the natural ligand EGF (8 nm final face of target cells. concentration). The results are shown in Figure 5c. Both Requirement of endosome escape activity for TEG- molecules efficiently competed with TEG for the binding mediated gene transfer to EGF receptor thereby drastically decreasing the The efficiency of TEG-mediated gene transfer was also expression of the luciferase reporter gene. These data investigated using human A431 tumor cells which carry Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 526 amplified copies of the EGF receptor gene and highly overexpress the receptor protein (2 million receptors per cell20). As shown in Figures 2 and 3, TEG binds to and activates EGF receptors on these cells. However, no efficient gene transfer could be obtained using a transfection protocol similar to that for COS-1 cells (Figure 6). In contrast, efficient luciferase reporter gene expression was observed after transfection of A431 cells in the presence of 100 ␮m chloroquine. Chloroquine has been shown to enhance the efficiency of various nonviral gene transfer vectors which utilize receptor-mediated endocytosis for DNA delivery.21–23 It blocks endosomal acidification, but also accumulates in intracellular vesicles and induces osmotic swelling of the endosomes which could result in endosome destabiliz- ation and the release of internalized DNA.24 Endosomal acidification is a process required for the activation of the exotoxin A translocation domain which serves as an endosome escape activity in carrier proteins such as TEG and the similar ErbB2-specific 5EG protein.6 Therefore, the dependence of TEG-mediated gene transfer on endosomal acidification was investigated in COS-1 cells using different inhibitors of endosomal acidification. The cells were transfected with TEG- pSV2G4LUC polyplex in the absence or presence of 100 ␮m chloroquine, 50 mm methylamine, 10 ␮m monensin, 5 ␮m nigericin, or 200 nm Bafilomycin A1. The results are shown in Figure 7. The addition of chloroquine as expected resulted in a 2.8-fold increase in reporter gene expression, whereas reagents which do not induce endosome disruption such as methylamine, monensin, niger- Figure 7 TEG-mediated gene transfer is dependent on the activity of the icin, and Bafilomycin A1, similar to previous results with translocation domain. COS-1 cells were treated with TEG-pSV2G4LUC 6,25 polyplex at a final concentration of refolded TEG of 4 nm for 4 h in the an ErbB2 specific carrier protein, led to a significant presence of (a) 100 ␮m chloroquine, or (b) 50 mm methylamine, 10 ␮m reduction in transfection efficiency. These data clearly monensin, 5 ␮m nigericin, or 200 nm Bafilomycin A1 as indicated. For show that in COS-1 cells endosomal acidification is Bafilomycin A1 the reagent was added to the cells 30 min before the trans- required for full gene delivery activity of the TEG carrier fection. Cells were harvested and reporter gene expression was analyzed protein and indicate that the exotoxin A translocation 36 h later. Relative luciferase activities in comparison to cells treated for the respective time intervals with TEG-pSV2G4LUC polyplex alone are shown.

domain of TEG is functionally active. This suggests that cell type dependent factors such as different intracellular routing of TEG–EGF receptor complexes and rapid receptor recycling might be responsible for the observed dependence on chloroquine of TEG-mediated gene transfer in A431 cells.

Discussion Overexpression of the EGF receptor has been found in many human tumors of epithelial origin and has been correlated with unfavorable patient prognosis.11 Since the EGF receptor is accessible on the cell surface and internalizes after ligand binding it is a suitable target for tumor-cell specific therapies which involve the intracellular delivery of macromolecules. This has been exploited Figure 6 TEG carrier protein-mediated gene transfer in A431 tumor cells. for recombinant toxins with antitumoral activity which A431 cells were seeded in 12-well tissue culture plates at a density of consist of natural EGF receptor ligands such as TGF-␣, 4 × 104 cells per well and grown overnight. Then the cells were treated or receptor-specific antibody domains fused to a bacterial with TEG-pSV2G4LUC polyplex (TEG) at a final concentration of toxin.16,19,20 Similarly, several strategies have been refolded TEG of 4 nm for 4 h as described in the legend to Figure 4 either ␮ developed to achieve the delivery of genetic material into in the presence or absence of 100 m chloroquine (Chl) as indicated. Con- EGF receptor-expressing cells. Direct targeting of retrovi- trol cells were incubated with pSV2G4LUC polyplex lacking the fusion protein (pL). Cells were harvested and reporter gene expression was ana- ral vectors has been approached by engineering the N- lyzed 36 h later. Luciferase activity is expressed in relative light units per terminus of the viral envelope protein via insertion of the milligram of total protein. natural ligand EGF.13 Likewise EGF receptor-mediated Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 527 delivery of nonviral vectors has been achieved using a demonstrated. Antibody–polylysine conjugates alone chemical conjugate of polylysine and an EGF receptor- were not very efficient in DNA delivery.12 Similarly, a specific antibody12 or a complex consisting of biotinyl- complex of EGF and polylysine connected via a biotin- ated EGF non-covalently bound to a streptavidin–polyly- streptavidin linkage did not result in efficient transfection sine conjugate.14 In the latter case the addition of adeno- of cells unless adenoviral particles were provided as an viral particles linked to polylysine was required to endosomolytic activity.14 Also retroviral vectors engine- facilitate efficient gene delivery into lung cancer cells. ered to target EGF receptor-expressing cells required the We have chosen the EGF receptor as a target for a addition of chloroquine for partial recovery of infec- modular fusion protein, TEG, which combines in a single tivity.13 Here the TEG fusion protein mediated efficient polypeptide chain DNA-binding, cell recognition, and gene delivery in COS-1 cells in the absence of additional endosome escape activities. The TEG protein is very simi- endosome disturbing reagents and the transfection lar in its structure to the previously described 5EG mol- efficiency of TEG-polyplex could be strongly reduced by ecule6 which facilitates cell-specific gene transfer and co-treatment of cells with different inhibitors of endoso- consists of a single chain antibody domain specific for the mal acidification. These results demonstrate that TEG, human ErbB2 receptor, an internal translocation domain like the parental exotoxin A, is dependent on the transit derived from Pseudomonas exotoxin A, and the sequence- through a low pH environment for full activity, and indi- specific DNA-binding domain of the yeast GAL4 protein. cate that the bacterial translocation domain contributes to To investigate whether this concept could be adapted for TEG-mediated gene delivery. In contrast, in EGF receptor distinct target cell populations, the N-terminal antibody overexpressing A431 cells TEG alone was unable to facili- domain of 5EG was replaced by the natural EGF receptor tate transfection, but for successful gene delivery was ligand TGF-␣ as a cell targeting domain. The resulting strictly dependent on the simultaneous presence of the TEG protein retained endosome escape and DNA-bind- endosome destabilizing reagent chloroquine. Similar ing activities but acquired a novel specificity for EGF results have also been obtained with EGF receptor- receptor-expressing cells. expressing HeLa human cervical carcinoma cells (data The TGF-␣ domain contains three disulfide bonds not shown). which upon expression in E. coli appear to be formed The ETA translocation domain is effective in delivering improperly resulting in decreased receptor binding the enzymatically active fragment of the wild-type toxin activity. Therefore an additional denaturation/renatur- to the cytosol where a few active molecules are sufficient ation step was included in the purification protocol to cause cell death.10 In the case of chimeric DNA carrier which greatly enhanced the activity of the TEG fusion proteins which utilize this bacterial domain, much bigger protein. Incubation of EGF receptor expressing COS-1 complexes must be released from the endosomes and cells with polyplex containing refolded TEG resulted in higher numbers of DNA molecules have to be delivered an up to 150-fold increase in reporter gene expression in in order to achieve substantial gene expression. Therefore comparison to polyplex alone. Thereby TEG-mediated it is not surprising that the addition of supplementary gene transfer could be blocked by the addition of EGF endosome destabilizing reagents can further improve or the antagonistic EGF receptor specific antibody 14E1 transfection efficiency.6 Likewise the intracellular deliv- indicating that DNA uptake in these cells is mediated via ery of a heterologous protease domain by ETA sequences specific binding of TEG-polyplex to the target receptor. was enhanced in the presence of adenoviral particles.27 Complexes of the fusion protein and DNA must be However, the dependence on chloroquine of TEG- internalized and reach an acidic environment before the mediated gene transfer in A431 cells was unexpected and ETA translocation domain can become activated and suggests that cell type specific factors strongly affect the facilitate the escape of the DNA to the cytosol. Binding intracellular delivery of the TEG-containing polyplex. of growth factors to the EGF receptor triggers receptor Due to enhanced receptor recycling in A431 cells,28 or dimerization followed by receptor activation and intern- altered intracellular routing, internalized TEG–EGF- alization of the ligand–receptor complex along the endo- receptor complexes might not reach an acidic environ- cytic pathway, thereby initiation of signal transduction ment or might not remain at low pH long enough to and internalization of the EGF receptor coincide.17 TGF- allow gene delivery to the cytosol. In addition, A431 cells ␣ is depleted from the medium at a faster rate than EGF26 respond abnormally to nanomolar concentrations of and might therefore be a better candidate as a ligand for exogenous growth factors. Natural EGF receptor ligands targeted gene delivery. Since the TEG fusion protein like usually stimulate cell proliferation while activation of natural growth factors induces autophosphorylation of EGF receptors in A431 cells can result in the suppression the EGF receptor it is very likely that it also triggers rapid of cell growth,29 most likely mediated via signaling internalization. A further argument to select TGF-␣ came through STAT proteins.30 We have observed that TEG from the observation that TGF-␣ itself might provide like TGF-␣ at concentrations which are nontoxic for COS- some endosome escape activity.19 Small chimeric toxins 1 cells induced cell death in A431 cells (data not shown). consisting only of the growth factor and ETA domain III While sublethal concentrations of TEG were used for the in the absence of the translocation domain already exhib- transfection experiments, it cannot be excluded that at ited high cytotoxic activity towards EGF receptor- least in this cell line continuing receptor activation by the expressing cells indicating that the toxin reached the TGF-␣ domain had an inhibitory effect on reporter gene cytosol. This additional activity of TGF-␣ might com- uptake and expression. Previously the ErbB2-specific 5EG plement the function of the internal translocation domain carrier protein which contains a recombinant antibody in the TEG fusion protein. fragment6 instead of a growth factor domain was shown The importance of an endosome escape activity for suc- to facilitate gene transfer in human tumor cells in the cessful uptake and intracellular delivery of protein–DNA absence of chloroquine. Similarly, the use of a DNA-bind- complexes via the EGF receptor has previously been ing fusion protein harboring an EGF receptor-specific Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 528 antibody domain16,20 instead of TGF-␣ might help to min at 4°C and loaded on to a Ni2+-saturated Chelating evaluate the influence of receptor activation on gene Sepharose Fast Flow column (Pharmacia Biotech, Frei- delivery further. burg, Germany) under denaturing conditions (8 m urea, We have shown that the concept of modular fusion 50 mm Tris-HCl, pH 8.0, 150 mm NaCl). Unbound pro- proteins for targeted gene delivery is versatile and allows teins were removed by washing the column with equili- the adaptation for different target cell populations by bration buffer containing 50 mm imidazole. Specifically including a suitable cell recognition domain. The TGF-␣- bound proteins were eluted by increasing the imidazole containing TEG DNA carrier protein is able to specifically concentration of the buffer to 250 mm. Fractions contain- recognize and transfer genetic material into EGF recep- ing the fusion protein were pooled, and the denaturant tor-expressing cells. Thereby cell type-specific factors was removed by dialysis against 50 mm Tris-HCl, pH 8.0, ␮ appear to greatly influence transfection efficiencies and 50 mm KCl, 5 mm MgCl2, 100 m ZnCl2, 20% glycerol, in have to be considered for individual applications of this the presence of 0.4 ml-arginine to avoid protein aggre- strategy. While the in vivo stability of protein–DNA com- gation. l-Arginine was then removed by dialysis directly plexes containing such modular fusion proteins remains before preparation of the TEG–DNA transfection com- to be established, their similarity to recombinant toxins plexes. A typical yield of purified TEG protein was 0.5 which have already demonstrated their capability to mg per liter of original bacterial culture with a purity of reach established tumors in vivo20,31 suggests that they approximately 90% determined by SDS-PAGE and might become useful reagents for the targeted delivery Coomassie brilliant blue staining. of therapeutic genes in cancer gene therapy. Alternatively, for complete denaturation and reduction of disulfide bonds, eluates from the Ni2+-affinity column were directly treated with 0.3 m DTT for 2 h at room Materials and methods temperature. Then the samples were rapidly diluted to a protein concentration below 100 ␮g/ml in refolding Cell lines buffer containing 100 mm Tris-HCl, pH 8.0, 8 mm oxid- COS-1 SV40-transformed African green monkey kidney ized glutathione, 0.5 ml-arginine, 150 mm NaCl. Refold- cells and A431 human epidermoid carcinoma cells were ing of the protein was allowed to take place for 2 days maintained in Dulbecco’s modified Eagle’s medium at 10°C. Subsequently the renatured protein was concen- (DMEM; BioWhittaker, Verviers, Belgium) supplemented trated and dialyzed as described above. with 10% heat inactivated fetal bovine serum (FBS), 2 mm l-glutamine, 100 units/ml penicillin and 100 ␮g/ml ELISA experiments ° streptomycin at 37 C in a humidified atmosphere of air The binding of TEG to the EGF receptor was measured and 5% CO2. by ELISA as described.16 Ninety-six-well microtiter plates coated with 100 ng/well of recombinant protein compris- Construction of the TEG fusion protein ing the extracellular domain of the human EGF receptor A DNA fragment encoding amino acids 1 to 50 of human were blocked with 3% BSA in TBS (10 mm Tris-HCl, pH TGF-␣, six consecutive histidine residues and amino 7.5, 150 mm NaCl). Fifty microliters of TEG at concen- acids 252 to 308 of exotoxin A (ETA) from Pseudomonas trations ranging from 0.05 to 333 nm were added to the aeruginosa was derived by HindIII/SalI digestion of the wells and the plates were incubated for 1 h at room tem- plasmid pSW202-TGF␣ which encodes a TGF-␣-ETA perature. Unbound protein was removed, the wells were fusion protein.31 A second DNA fragment encoding washed and incubated with MAb RK5C1 specific for the amino acids 309 to 366 of ETA fused to amino acids 2 to GAL4 DNA-binding domain (Santa Cruz Biotechnology, 147 of the yeast transcriptional activator GAL4 was iso- Heidelberg, Germany) for 1 h at room temperature fol- lated from the plasmid pSW50–5EG6 by SalI/XbaI diges- lowed by incubation with goat anti-mouse IgG coupled tion. The HindIII/SalI and the SalI/XbaI fragments were to alkaline phosphatase (Sigma, Deisenhofen, Germany). inserted into a HindIII/XbaI digested pFLAG-1 vector Specifically bound protein was detected by incubation (Integra Biosciences, Fernwald, Germany). The resulting with a solution of 100 mm glycine, 1 mm ZnCl2,1mm plasmid pWF47-TEG encodes under the control of the MgCl2, pH 10.4, 1 mg/ml p-nitrophenylphosphate diso- IPTG-inducible tac promoter the TEG fusion protein dium (Sigma) for 3 h at room temperature, then the which consists of an E. coli ompA signal peptide, the syn- absorbance at 405 nm was measured. thetic FLAG epitope encoded by the vector, the human TGF-␣ domain, a cluster of six His residues, the ETA FACS analysis translocation domain, and the GAL4 DNA-binding A431 cells (5 × 105) in PBS were incubated with 1.6 ␮gof domain. TEG in the presence or absence of 1 ␮g of EGF (four-fold molar excess) for 40 min at 4°C. Unbound proteins were Bacterial expression and purification of the TEG fusion removed, cells were washed twice with PBS, and then protein treated with 3 ␮g of the anti-FLAG MAb M2 (Integra E. coli CC118 carrying the plasmid pWF47-TEG were Biosciences) for another 40 min followed by FITC-labeled ° grown at 37 CtoanOD550 of 0.5 in LB medium contain- goat anti-mouse IgG as a second antibody. Fluorescence ing 0.6% glucose and 100 ␮g/ml ampicillin. Protein of cells was analyzed with a fluorescence-activated cell expression was induced for 1 h at 37°C by the addition sorter (FACScan; Becton Dickinson, Heidelberg, of 0.5 mm isopropyl-␤-d-thiogalactopyranoside (IPTG). Germany). Cells were harvested by centrifugation, resuspended in 50 mm Tris-HCl, pH 8.0, containing 8 m urea, 150 mm EGF receptor activation assays NaCl and 0.3 mm PMSF, and lysed by sonication. Cell COS-1 and A431 cells were grown for 8 h in DMEM con- lysates were cleared by centrifugation at 30 000 g for 30 taining 0.5% FCS. After removal of the medium, purified Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 529 TEG was added at a concentration of 3 ␮g/ml in PBS for Acknowledgements 10 min at 37°C in the presence or absence of the antagonistic EGF-receptor specific monoclonal antibody 14E116 as We thank Dr M Schmidt for helpful suggestions and for a competitor. Control cells were treated with PBS or 100 providing plasmid pSW202-TGF␣, Dr D Schneider for Mab 14E1 and purified EGF receptor protein, and Dr R ng/ml EGF. Cell lysis was performed in a buffer contain- ¨ ing 50 mm Tris-HCl, pH 7.5, 5 mm EDTA, 150 mm NaCl, Ruger, Boehringer Mannheim GmbH, for support of this work. 1% Triton X-100, 2 mm NaVO4,50mm NaF and 10 mm Na-molybdate. Cleared cell lysates containing equal amounts of total proteins were separated by SDS-PAGE on a 7.5% gel and subsequently blotted on a PVDF mem- References brane (Millipore, Eschborn, Germany). Phosphotyrosine- 1 Ledley FD. Nonviral gene therapy: the promise of genes as containing proteins were detected with an antiphospho- pharmaceutical products. Hum Gene Ther 1995; 6: 1129–1144. tyrosine monoclonal antibody (Santa Cruz Bio- 2 Cooper MJ. Noninfectious gene transfer and expression systems technology), followed by incubation with an antimouse for cancer gene therapy. Semin Oncol 1996; 23: 172–187. horseradish peroxidase coupled antibody and chemilum- 3 Perales JC, Ferkol T, Molas M, Hanson RW. An evaluation of inescent detection with the ECL kit (Amersham, receptor-mediated gene transfer using synthetic DNA–ligand Braunschweig, Germany). The amount of EGF receptors complexes. Eur J Biochem 1994; 226: 255–266. 4 Cristiano RJ, Curiel DT. Strategies to accomplish gene delivery in each lane was analyzed by reincubation of the mem- via the receptor-mediated endocytosis pathway. Cancer Gene brane with the EGF-receptor specific 12E rabbit anti- Ther 1996; 3: 49–57. 32 serum. 5 Chen SY, Zani C, Khouri Y, Marasco WA. Design of a genetic immunotoxin to eliminate toxin immunogenicity. Gene Therapy 1995; 2: 116–123. Transfection of cells with TEG-pSV2G4LUC polyplex 6 Fominaya J, Wels W. Target cell-specific DNA transfer mediated Transfection complexes were prepared by incubation of by a chimeric multidomain protein. Novel non-viral gene deliv- 2 ␮g pSV2G4LUC plasmid DNA6 with variable amounts ery system. J Biol Chem 1996; 271: 10560–10568. 7 Paul RW et al. Gene transfer using a novel fusion protein, of the TEG fusion protein for 15 min at room temperature GAL4/invasin. Hum Gene Ther 1997; 8: 1253–1262. in a buffer containing 50 mm Hepes, pH 7.5, 0.1 mg/ml 8 Allured VS, Collier RJ, Carroll SF, McKay DB. Structure of ␮ BSA, 50 mm KCl, 5 mm MgCl2, 100 m ZnCl2. To facilitate exotoxin A of Pseudomonas aeruginosa at 3.0-Angstrom resol- charge neutralization and condensation of the protein– ution. Proc Natl Acad Sci USA 1986; 83: 1320–1324. DNA complex, 1.25 ␮g of poly-l-lysine HBr (Sigma) with 9 Hwang J, Fitzgerald DJ, Adhya S, Pastan I. Functional domains an average degree of polymerization of 236 residues of Pseudomonas exotoxin identified by deletion analysis of the (pL ) were added slowly to the complex (pL :DNA gene expressed in E. coli. Cell 1987; 48: 129–136. 236 236 10 Ogata M, Fryling CM, Pastan I, FitzGerald DJ. Cell-mediated molar ratio of 50:1) and the mixture was incubated for cleavage of Pseudomonas exotoxin between Arg279 and Gly280 an additional 15 min. For transfection, cells were seeded generates the enzymatically active fragment which translocates 4 in 12-well tissue culture plates at a density of 4 × 10 cells to the cytosol. J Biol Chem 1992; 267: 25396–25401. per well and grown overnight at 37°C. Growth medium 11 Gullick WJ. Prevalence of aberrant expression of the epidermal was exchanged with 1 ml per well of fresh medium 5 h growth factor receptor in human cancers. Br Med Bull 1991; 47: before the addition of 100 ␮l of protein–DNA complexes. 87–98. The cells were incubated for 16 h at 37°C with TEG-poly- 12 Chen J, Gamou S, Takayanagi A, Shimizu N. A novel gene delivery system using EGF receptor-mediated endocytosis. FEBS Lett plex at a final DNA concentration of 0.5 nm, a final pL236 1994; 338: 167–169. concentration of 25 nm, and the final TEG concentration 13 Cosset FL et al. Retroviral retargeting by envelopes expressing as indicated in the respective Figure legends, then the an N-terminal binding domain. J Virol 1995; 69: 6314–6322. medium was exchanged and the cells were grown for 14 Cristiano RJ, Roth JA. Epidermal growth factor mediated DNA another 36 h before they were harvested for analysis. delivery into lung cancer cells via the epidermal growth factor receptor. Cancer Gene Ther 1996; 3: 4–10. 15 Seetharam S, Chaudhary VK, FitzGerald D, Pastan I. Increased Luciferase assay cytotoxic activity of Pseudomonas exotoxin and two chimeric toxins ending in KDEL. J Biol Chem 1991; 266: 17376–17381. The growth medium was removed, the cells were washed 16 Schmidt M, Vakalopoulou E, Schneider DW, Wels W. Construc- twice with PBS, and lysed for 15 min at room tempera- tion and functional characterization of scFv(14E1)-ETA – a ture in 100 ␮l of buffer containing 25 mm glycylglycine, novel, highly potent antibody-toxin specific for the EGF recep- pH 7.8, 1 mm DTT, 15% glycerol, 8 mm MgSO4,1mm tor. Br J Cancer 1997; 75: 1575–1584. EDTA, and 1% Triton X-100. The lysates were cleared by 17 Bergeron JJ et al. Endosomes, receptor tyrosine kinase internaliz- centrifugation and protein content was determined. Fifty ation and signal transduction. Biosci Rep 1995; 15: 411–418. microliters of the lysate were mixed with an identical vol- 18 Felgner PL et al. Nomenclature for synthetic gene delivery systems. Hum Gene Ther 1997; 8: 511–512 (editorial). ume of dilution buffer containing 25 mm glycylglycine, 19 Kihara A, Pastan I. Small chimeric toxins containing only trans- pH 7.8, 10 mm MgSO4 and 5 mm ATP. Luciferase activity forming growth factor alpha and domain III of Pseudomonas was monitored for 30 s in a luminometer AutoLumat LB exotoxin with good antitumor activity in mice. Cancer Res 1994; 953 (Berthold, Munich, Germany) with automatic injec- 54: 5154–5159. tion of 300 ␮l luciferin solution containing 250 mm lucif- 20 Wels W et al. EGF receptor and p185erbB-2-specific single-chain erin (Sigma), 25 mm glycylglycine, pH 7.8 and 0.5 mm antibody toxins differ in their cell-killing activity on tumor cells coenzyme A (Boehringer Mannheim, Mannheim, expressing both receptor proteins. Int J Cancer 1995; 60: 137–144. Germany). Luciferase activity was determined as relative 21 Zenke M et al. Receptor-mediated endocytosis of transferrin– polycation conjugates: an efficient way to introduce DNA into light units per milligram of cellular protein. hematopoietic cells. Proc Natl Acad Sci USA 1990; 87: 3655–3659. Cell-specific gene transfer mediated by a TGF-␣ fusion protein J Fominaya et al 530 22 Cotten M et al. Transferrin–polycation-mediated introduction of 27 Novoa I, Cotten M, Carrasco L. Hybrid proteins between Pseudo- DNA into human leukemic cells: stimulation by agents that monas aeruginosa exotoxin A and poliovirus 2Apro cleave p220 affect the survival of transfected DNA or modulate transferrin in HeLa cells. J Virol 1996; 70: 3319–3324. receptor levels. Proc Natl Acad Sci USA 1990; 87: 4033–4037. 28 Masui H, Castro L, Mendelsohn J. Consumption of EGF by A431 23 Fritz JD, Herweijer H, Zhang G, Wolff JA. Gene transfer into cells: evidence for receptor recycling. J Cell Biol 1993; 120: 85–93. mammalian cells using histone-condensed plasmid DNA. Hum 29 Gill GN, Lazar CS. Increased phosphotyrosine content and inhi- Gene Ther 1996; 7: 1395–1404. bition of proliferation in EGF-treated A431 cells. Nature 1981; 24 Wagner E, Curiel D, Cotten M. Delivery of drugs, proteins and 293: 305–307. genes into cells using transferrin as a ligand for receptor- 30 Chin YE et al. Cell growth arrest and induction of cyclin-depen- mediated endocytosis. Adv Drug Del Rev 1994; 14: 113–135. dent kinase inhibitor p21 WAF1/CIP1 mediated by STAT1. 25 Wels W, Fominaya J. Peptides and fusion proteins as modular Science 1996; 272: 719–722. DNA carriers. In: Kabanov AV, Felgner P, Seymour LW (eds). 31 Schmidt M, Wels W. Targeted inhibition of tumour cell growth Self-Assembling Complexes for Gene Delivery From Laboratory to by a bispecific single-chain toxin containing an antibody domain Clinical Trial. Wiley: Chichester, 1998, pp 351–369. and TGF alpha. Br J Cancer 1996; 74: 853–862. 26 Reddy CC, Wells A, Lauffenburger DA. Receptor-mediated 32 Gullick WJ, Downward J, Waterfield MD. Antibodies to the effects on ligand availability influence relative mitogenic autophosphorylation sites of the epidermal growth factor recep- potencies of epidermal growth factor and transforming growth tor protein-tyrosine kinase as probes of structure and function. factor alpha. J Cell Physiol 1996; 166: 512–522. EMBO J 1985; 4: 2869–2877.