Abstract
Background: In this study, different variants of the anti-PSMA single-chain antibody fragment (scFv) D7 were cloned, varying in linker type, position of hexahistidine tags and VH-VL orientation. From these scFv, Pseudomonas exotoxin A-based immunotoxins were constructed and their biological activities against prostate cancer cells were compared. Materials and Methods: Binding of the constructs to PSMA-expressing prostate cancer cells was determined by flow cytometry. ADP-ribosyltransferase activity was analysed and cytotoxicity was measured with WST-1 assays. Results: Different linker types did not influence the characteristics of the scFv or immunotoxins. The addition of an N-terminal hexahistidine-tag, however, resulted in decreased expression, binding and cytotoxicity. scFv in VH-VL orientation showed the highest expression and binding, whereas immunotoxins in VL-VH orientation exhibited the best binding and cytotoxicity. Conclusion: The present study showed how structure influences the characteristics of scFv and immunotoxins. It is therefore suggested that for each individual construct the optimal structure has to be determined separately.
Antibodies are useful molecules for targeted diagnostic and therapeutic applications in cancer research since they are unique in their high affinity and specificity for certain tumour antigens. The variable region (Fv) portion of an antibody comprises the variable domains of the heavy (VH) and of the light chain (VL). It represents the smallest antibody constituent containing a complete antigen binding site. Single chain antibody fragments (scFv) are recombinant proteins that contain the VH and VL of an antibody, connected by a flexible peptide linker. They are appreciated as diagnostic and therapeutic alternatives, due to their small size. This facilitates the penetration of tumour tissues, improves the pharmacokinetics and reduces their immunogenicity in comparison to mAbs or Fab fragments (1-4). Furthermore, the recombinant production of scFv renders genetic manipulation possible and the direct attachment of functional domains. This improves the stability of the scFv and alleviates their production, therefore being advantageous compared to chemical coupling approaches. Several properties of an scFv, including its protein sequence (5, 6), the order of its variable domains (7, 8), the linker type (9), the position of functional tags for purification and detection (10) and the attachment of functional domains (11, 12), have been shown to influence the scFv expression yield and antigen binding.
In recent years, prostate-specific membrane antigen (PSMA), a type II membrane glycoprotein, has generated intense interest as a target for therapeutic intervention against prostate cancer, since it is abundantly expressed in prostate cancer cells correlating with tumour progression (13, 14). Moreover, PSMA is expressed on the cell surface without being released into the circulation and internalises upon antibody binding (15). Different antibodies against PSMA have been developed, of which the most promising for a targeted therapy and diagnosis are those recognising the three-dimensional structure of PSMA on the cell surface (16, 17).
An anti-PSMA scFv, called D7, was generated from the monoclonal antibody 3/F11, which shows an excellent and specific binding to cell-adherent PSMA (12). Additionally, an immunotoxin was generated by C-terminal ligation of the toxic domain of Pseudomonas exotoxin A (PE40) to D7. This molecule, called D7-PE40, showed a high cytotoxicity against prostate cancer cells in vitro and caused a significant inhibition of tumour growth in vivo (12).
Overview of different single-chain antibody fragments (scFv) and immunotoxins in the expression vector pHOG21 used in this study. VH, variable domain of the heavy chain; VL, variable domain of the light chain; Yol, Yol epitope (EEGEFSEAR); GS, peptide linker (G4S)3; c-MYC, human c-MYC tag (EQKLISEEDL); His6, hexahistidine tag (HHHHHH); PE40, toxic fragment of Pseudomonas exotoxin A.
The present study examined the influence of the VH-VL orientation, the linker type and the position of hexahistidine (His6)-tags on the expression and biological activities of different variants of the scFv D7 and the immunotoxin D7-PE40.
Materials and Methods
Cell lines. For testing the binding capacities of the scFv and their corresponding immunotoxins, the PSMA-positive prostate cancer cell line C4-2 was used. The cells were cultured in RPMI-1640 medium (Gibco Invitrogen, Karlsruhe, Germany), supplemented with penicillin (100 U/ml), streptomycin (100 mg/l) and 10% foetal calf serum (Biochrom, Berlin, Germany) at 37°C in a humidified atmosphere of 5% CO2.
Cloning of the D7-based anti PSMA scFv and immunotoxins. The generation of the anti-PSMA scFv D7 has been described previously (12). In brief, the V regions of D7 were selected by phage display from the anti-PSMA hybridoma 3/F11. The gene sequence of the VH of D7 was then cloned between the NcoI and HindIII restriction sites and the VL of D7 between the MluI and NotI restriction sites into the vector pHOG21 (18). This vector contained a hydrophobic pel-B leader for periplasmatic expression, the 9 aa Yol-epitope as peptide linker between the VH and VL domains and C-terminal sequences for a human c myc tag for protein detection and a His6-tag for purification. In the present study, this scFv was termed D7-VH(Yol)VL and used as the starting construct (Figure 1A).
For exchanging the Yol-linker against a (G4S)3-linker, the double-stranded gene sequence of the (G4S)3-linker containing an N-terminal HindIII restriction site and a C-terminal MluI restriction site was synthesised (GeneArt, Regensburg, Germany) and inserted into the scFv D7 VH(Yol)VL. The resulting scFv was termed D7-VH(GS)VL.
An additional His6-tag was attached at the N-terminus of the scFv D7-VH(GS)VL by PCR, using a forward primer containing a His6-tag. This construct was termed His-D7-VH(GS)VL.
To generate scFv in a VL-VH orientation, the VH domain was amplified by PCR, using a forward primer containing MluI restriction site and a reverse primer containing NotI restriction site. For the VL domain, a forward primer containing NcoI restriction site and a reverse primer containing HindIII restriction site were used. The VL and VH were then inserted, via the NcoI/HindIII and the MluI/NotI restriction sites, respectively, into the scFv D7-VH(GS)VL, resulting in the scFv D7-VL(GS)VH.
An N-terminal His6-tag was added to the scFv D7-VL(GS)VH by PCR, using a forward primer containing the sequence of the His6-tag. This resulted in the scFv His-D7-VL(GS)VH.
The gene sequence encoding the truncated form of Pseudomonas exotoxin A (PE40, aa 252-613) with ADP-ribosyltransferase activity was amplified by PCR from the vector pSW200. It was C-terminally ligated to the sequences of the scFv, lacking their stop codons via the XbaI restriction site of the vector pHOG21. According to the names of their scFv domains, the immunotoxins were termed D7-VH(Yol)VL-PE40, D7-VH(GS)VL-PE40, His D7-VH(GS)VL-PE40, D7-VL(GS)VH-PE40 and His-D7-VL(GS)VH-PE40, respectively (Figure 1B). All scFv and immunotoxin constructs were confirmed by sequencing.
Periplasmatic expression of the scFv and immunotoxins. The D7-based scFv and immunotoxins were periplasmatically expressed in Escherichia coli XL1-blue cells (Stratagene, La Jolla, CA, USA) and purified by immobilised metal affinity chromatography (IMAC). For this, bacteria were transformed with the pHOG21 plasmids and grown overnight in 2× YT medium, diluted 1:20 and grown again in a volume of 600 ml at 37°C. After reaching an optical density (OD) of 0.8 A/cm at 600 nm, the bacteria were pelleted by centrifugation and resuspended in 600 ml of fresh 2× YT medium with 50 μg/ml ampicillin, 0.4 M sucrose and 1 mM IPTG. The culture was incubated for 18 h at room temperature. After centrifugation, the bacteria were resuspended in 24 ml lysis buffer (50 mM NaH2PO4, 300 mM NaCl, pH 8.0) and incubated for 30 min on ice. The periplasmatically expressed proteins were obtained in the supernatant after another centrifugation step at 14,000 ×g. For the purification by IMAC, 1.5 ml of Ni2+-charged agarose (Qiagen, Hilden, Germany) were added to the periplasmatic extract, followed by incubation on a shaker for 1 h at 4°C. The mixture was loaded onto a polypropylene column with subsequent sedimentation of the protein-bound agarose and flow-through of unbound protein. After this, the agarose was washed twice with buffer containing 50 mM NaH2PO4, 300 mM NaCl, and 20 mM imidazole (pH 8.0). Finally, the agarose-bound protein was eluted with the same buffer containing 250 mM imidazole. Eluted material was dialysed against PBS. Determination of the protein content was performed with the Micro BCA Protein Reagent Kit (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer's protocol.
Western blots of the recombinant scFv (30 kDa) and immunotoxins (70 kDa). The proteins were detected by a peroxidase-labelled mouse anti-human c-MYC mAb and the blot was developed with 3,3′-diaminobenzidine as substrate. (A) scFv: lane 1, D7-VH(Yol)VL; lane 2, D7 VH(GS)VL; lane 3, His-D7-VH(GS)VL; lane 4, D7-VL(GS)VH; lane 5, His-D7-VL(GS)VH. (B) Immunotoxins: lane 1, D7-VH(Yol)VL-PE40; lane 2, D7-VH(GS)VL-PE40; lane 3, His-D7-VH(GS)VL-PE40; lane 4, D7-VL(GS)VH-PE40; lane 5, His-D7-VL(GS)VH-PE40; M, protein marker.
Western blot analysis. The purified scFv and immunotoxins were analysed by SDS-PAGE and Western blotting according to the manufacturer's protocols (Invitrogen, Carlsbad, CA, USA). For Western blotting, the proteins were transferred onto nitrocellulose membranes after gel electrophoresis. After blocking with 5% low-fat milk in PBS-Tween 20 (0.05% v/v), the recombinant scFv and immunotoxins were detected by a peroxidase-labelled mouse anti-human c-MYC mAb (Roche Diagnostics, Mannheim, Germany). After washing extensively with PBS-Tween 20, the blot was developed with 3,3′-diaminobenzidine as substrate.
Flow cytometry. The cell binding affinities of the scFv and immunotoxins were evaluated by flow cytometry. For this method, 2×105 C4-2 cells in PBS, containing 3% FCS and 0.1% NaN3, were incubated with different concentrations of each scFv or immunotoxin for 1 h on ice. Following three rounds of washing with PBS, cells were incubated with a mouse anti-human c-MYC mAb (Roche Diagnostics) for 40 min on ice. After three further washing steps, the cells were incubated with goat anti-mouse Ig-R-PE (Becton Dickinson, Mountain View, CA, USA) for 30 min on ice. Afterwards, the cells were washed again and resuspended in 200 μl PBS containing 1 μg/ml propidium iodide. Mean fluorescence intensities of the stained cells were measured and analysed using a FACScan flow cytometer and the CellQuest Pro Software (BD Biosciences, Heidelberg, Germany).
ADP-ribosyltransferase assay. For determining the ADP-ribosyltransferase activity of the immunotoxins, a Western blot, detecting the ADP-ribosylation of eEF-2 by the PE40 domain, was performed according to a modified method published by Bachran et al. (19). For this, 0.5 μg immunotoxin, 0.5 μl 6 Biotin-17-NAD (250 μM; Biozol, Eching, Germany), 3 μl yeast eEF-2 (350 ng/μl), and 2.5 μl BSA (2 mg/ml) in a total volume of 25 μl reaction buffer (50 mM Tris, pH 7.6, 1 mM EDTA, 1 mM dithiothreitol) were incubated for 1 h at 37°C. Pseudomonas exotoxin A (Sigma Aldrich, Hamburg, Germany) served as a positive control.
The transfer of ADP-ribose from 6-Biotin-17-NAD to eEF-2 was confirmed by blotting the mixture and detection of the ADP-ribosylated eEF-2 using a goat anti-biotin-HRP antibody (Abcam, Cambridge, UK).
In vitro cytotoxicity. The in vitro cytotoxicity of the immunotoxins was determined using the WST-1 cell viability assay (Roche Diagnostics). For this, 1.5×104 C4-2 cells/well were seeded in a 96-well plate and incubated overnight. The immunotoxins were then added at different concentrations ranging from 10 pM to 25 nM. Afterwards, the cells were incubated for 48 h at 37°C and 5% CO2, followed by the addition of WST-1 reaction solution (15 μl/well) and further incubation for 90 min. The spectrophotometrical absorbances of the samples were measured at 450 nm (reference wavelength 690 nm). The IC50 value was defined as the concentration of immunotoxin inducing a 50% reduction in cell viability relative to untreated cells.
Periplasmatic expression yields and cell binding of the scFv and immunotoxins.
Results
Periplasmatic expression of the scFv and immunotoxins. The scFv and the immunotoxins were periplasmatically expressed in E. coli XL1-blue bacteria and purified by IMAC. Western blot analyses showed a molecular weight of approximately 30 kDa for the scFv (Figure 2A) and 70 kDa for the immunotoxins (Figure 2B).
The mean periplasmatic expression yield in one litre of bacterial culture ranged from 101.2 to 361.2 μg/l for the scFv group (Table I). The yields for D7-VH(Yol)VL (306.6 μg/l) and D7-VH(GS)VL (361.2 μg/l) were higher compared to the construct with the VL-VH constellation, D7-VL(GS)VH (205.2 μg/l). ScFv with an additional N-terminal His6-tag had the lowest yields (101.2 μg/l for His-D7-VH(GS)VL and 116.4 μg/l for His-D7-VL(GS)VH).
Compared to the scFv, higher expression yields were obtained in the immunotoxin group ranging from 246.2 to 596.1 μg/l (Table I). Similar to the scFv group, N-terminal His6-tagged constructs showed the lowest expression yields (275.1 μg/l for His-D7-VH(GS)VL-PE40 and 246.2 μg/l for His-D7-VL(GS)VH-PE40). The immunotoxin D7-VL(GS)VH-PE40 showed the highest yield with an average value of 596.1 μg/l.
Binding of the scFv and immunotoxins. The binding of the different scFv and immunotoxins was evaluated on PSMA-positive cells of the androgen-independent growing prostate cancer line C4-2. The binding properties were determined by flow cytometric analyses with increasing concentrations of scFv or immunotoxin, each derived from three individual bacterial inductions. For a comparison of the different constructs, the half-maximal saturation concentrations (HMSCs), corresponding to concentrations of scFv or immunotoxin reaching a half-maximal saturation of PSMA sites on C4-2 cells, were determined (Table I).
Cytotoxicity of the immunotoxins.
In the scFv group, the two constructs with the VH-VL constellation exhibited a somewhat higher binding capacity (HMSCs of 34.0 nM for D7-VH(Yol)VL and 24.0 nM for D7-VH(GS)VL) compared to the construct with VL-VH constellation (HMSC of 45.7 nM for D7-VL(GS)VH). The scFv with an N-terminal His6-tag showed lower binding capacities with HMSCs of 42.7 nM for His-D7-VH(GS)VL and 39.3 nM for His-D7-VL(GS)VH.
In contrast, in the immunotoxin group, the construct with the VL-VH orientation, D7 VL(GS)VH-PE40, exhibited the best cell binding. The HMSC was 18.3 nM, while the other immunotoxins had HMSCs between 44.0 and 63.3 nM (Table I).
ADP-ribosyltransferase activity. The PE40 domain has been widely used for the construction of immunotoxins against a broad panel of tumours. It ADP-ribosylates the eukaryotic elongation factor 2 (eEF-2), which ultimately leads to the inhibition of protein biosynthesis and the death of the target cells (20).
To investigate whether the conformations of the different immunotoxins affected the enzymatic activity of their PE40 domains, ADP-ribosyltransferase assays were performed. As demonstrated in Figure 3, all immunotoxins equally retained their ADP-ribosyltransferase activity. This indicates that changes in the conformation of the immunotoxins do not lead to a measurable loss of enzymatic activity.
ADP-ribosyltransferase activity of the immunotoxins was determined by Western blotting detecting the transfer of ADP-ribose from 6-biotin-17-NAD to eEF-2 (100 kDa) with a goat anti-biotin-HRP antibody. Lane 1, D7-VH(Yol)VL-PE40; lane 2, D7-VH(GS)VL-PE40; lane 3, His-D7-VH(GS)VL-PE40; lane 4, D7-VL(GS)VH-PE40; lane 5, His-D7-VL(GS)VH-PE40; lane 6, positive control with Pseudomonas exotoxin A; lane 7, negative control without immunotoxin; M, protein marker.
Cytotoxicity of the immunotoxins. The WST-1 viability assay was used for determining the ability of the different immunotoxins to inhibit the growth of PSMA-positive C4-2 cells (Table II). The immunotoxin D7-VL(GS)VH-PE40, which showed the best binding activity of all constructs, also displayed the highest cytotoxicity, with a mean IC50 value of 36.7 pM (Table II). The IC50 values of the other constructs were: 81.7 pM for D7-VH(GS)VL-PE40, 100 pM for His-D7-VH(GS)VL-PE40, 115.0 pM for His-VL(GS)VH-PE40 and 125.0 pM for D7-VH(Yol)VL-PE40.
Discussion
The scFv and immuntoxins of the present study were expressed in E. coli using the expression vector pHOG21 (18). This vector contains the hydrophobic pelB-leader sequence, which directs the constructs to be transported into the periplasm. There, the biochemical milieu promotes correct folding and formation of intrachain disulfide bonds. Although the scFv as well as the immunotoxins were also found to some extent in the cytoplasm and inclusion bodies (data not shown), only the periplasmatic fractions were used.
To connect the variable domains of the scFv, the vector pHOG21 contained the 9 aa Yol-epitope as a peptide linker that includes a tubulin epitope recognised by the monoclonal antibody Yol1/34 (21). As it is expected that the tubulin epitope may be immunogenic in patients, the Yol-epitope was exchanged with the widely used (G4S)3-linker. This linker, composed of glycines lacking side chains and serines with maximum hydrophilicity, is known to have a high flexibility and does not interfere with folding of the VH and VL domains (22). In the present study, the scFv D7-VH(Yol)VL showed similar expression yields and binding properties to its counterpart D7-VH(GS)VL containing the (G4S)3-linker. Comparable results were obtained with the corresponding immunotoxins D7-VH(Yol)VL-PE40 and D7-VH(GS)VL-PE40. This provided an indication that the Yol-epitope does not interfere with the variable domains of D7 and that its length and flexibility are sufficient for folding.
The use of affinity tags is preferred for the purification of recombinant proteins such as scFv and immunotoxins. However, the decision regarding the relative positioning of the affinity tags remains difficult and depends on the primary sequence and conformation of the protein. Cloning an additional His6-tag to the N-terminus of D7 resulted in reduced expression of the scFv in the VH-VL, as well as the VL-VH conformation. Additionally, similar or reduced binding activities were measured. This was in accordance with the N-terminal addition of His6-tags in the immunotoxin group, which reduced expression, cell binding and cytotoxicity. It is therefore suggested that the N-terminal His6-tag interferes with the D7 antigen-binding site. This is in contrast to a study by Goel et al., who demonstrated a covering of the antigen binding site of a pan-carcinoma scFv in VL-VH orientation by a C-terminal, but not by an N-terminal His6-tag (10).
Changing the orientation of the variable domains can affect the expression and activity of a scFv or immunotoxin. For instance, Hamilton et al. constructed a scFv directed against a melanoma-associated proteoglycan, showing a 1.5-fold elevated expression in VL-VH orientation compared to the VH-VL variant (23). In contrast, expression of an anti-c-Met scFv in the cytoplasm of E. coli was about five-fold higher in the VH-VL than in the VL-VH orientation (24). Francisco et al. expressed a PE40-based anti-CD40 immunotoxin in inclusion bodies and found that the binding and cytotoxicity of the construct G28 5sFv(VL VH)-PE40 was 10-fold higher compared to the VH-VL counterpart (25). In the present study, the order of the variable domains of D7 was also changed and the VH-VL scFv showed a somewhat higher expression and binding compared to the VL-VH constructs. In contrast, a very high expression, binding and cytotoxic activity of the immunotoxin was observed in VL-VH orientation. Therefore, for the use of D7 in a PE40-based immunotoxin, the VL-VH orientation should be preferred.
The anti-PSMA scFv D7 and its corresponding PE40-based immunotoxins represent interesting examples of how structural variations may affect expression and binding affinity. Therefore, different combinations should be investigated for each scFv in order to improve the biological activity of scFv-based constructs in the fight against cancer.
Acknowledgements
The present study was performed in the laboratory of Professor U. Elsässer-Beile. The Authors thank Professor M. Little for supplying the vector pHOG21, Professor W. Wels for providing the PE40 cDNA, and Professor H. Fuchs and Dr. C. Bachran for supplying yeast eEF-2. The work was supported by a grant of the Deutsche Krebshilfe e.V.
- Received June 4, 2010.
- Revision received June 23, 2010.
- Accepted July 1, 2010.
- Copyright© 2010 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved