Antibody internalization studied using a novel IgG binding toxin fusion

Journal of Immunological Methods 321 (2007) 41 – 59 www.elsevier.com/locate/jim

Research paper

Antibody internalization studied using a novel IgG binding toxin fusion Yariv Mazor a,1 , Itay Barnea b , Iafa Keydar b , Itai Benhar a,⁎ a

Department of Molecular Microbiology and Biotechnology, The George S. Wise Faculty of Life Sciences, Green Building, Room 202, Tel-Aviv University, Ramat Aviv 69978, Israel b Department of Cell Research and Immunology, The George S. Wise Faculty of Life Sciences, Tel-Aviv University, Ramat Aviv 69978, Israel Received 21 October 2006; received in revised form 29 November 2006; accepted 7 January 2007 Available online 6 February 2007

Abstract Targeted therapy encompasses a wide variety of different strategies, which can be divided into direct or indirect approaches. Direct approaches target tumor-associated antigens by monoclonal antibodies (mAbs) binding to the relevant antigens or by smallmolecule drugs that interfere with these proteins. Indirect approaches rely on tumor-associated antigens expressed on the cell surface with antibody–drug conjugates or antibody-based fusion proteins containing different kinds of effector molecules. To deliver a lethal cargo into tumor cells, the targeting antibodies should efficiently internalize into the cells. Similarly, to qualify as targets for such drugs newly-discovered cell-surface molecules should facilitate the internalization of antibodies that bind to them. Internalization can be studied be several biochemical and microscopy approaches. An undisputed proof of internalization can be provided by the ability of an antibody to specifically deliver a drug into the target cells and kill it. We present a novel IgG binding toxin fusion, ZZ-PE38, in which the Fc-binding ZZ domain, derived from Streptococcal protein A, is linked to a truncated Pseudomonas exotoxin A, the preparation of complexes between ZZ-PE38 and IgGs that bind tumor cells and the specific cytotoxicity of such immunocomplexes is reported. Our results suggest that ZZ-PE38 could prove to be an invaluable tool for the evaluation of the suitability potential of antibodies and their cognate cell-surface antigens to be targeted by immunotherapeutics based on armed antibodies that require internalization. © 2007 Elsevier B.V. All rights reserved. Keywords: Targeted therapy; Fusion protein; Internalization; ZZ domain; PE38; Immunotoxin

Abbreviations: CDR, complementarity determining region; DMF, Dimethyl Formamide; ELISA, enzyme-linked immunosorbent assay; FCS, fetal calf serum; FITC, fluorescein isothiocyanate; mAb, monoclonal antibody; FPLC, fast protein liquid chromatography; HRP, horseradish peroxidase; PBS, phosphate-buffered saline; PBST, PBS containing 0.05% Tween 20; PCR, polymerase chain reaction; PE, Pseudomonas exotoxin; PE38, a truncated Pseudomonas exotoxin lacking domain I and part of domain Ib.; RT, room temperature (25 °C); scFv, single chain antibody fragment; TBS, Tris-buffered saline; VH, heavy chain variable region; VL, light chain variable region; VNTR, the variable-number-of-tandem-repeat region of the MUC1 protein. ⁎ Corresponding author. Tel.: +972 3 6407511; fax: +972 3 6409407. E-mail address: [email protected] (I. Benhar). 1 Present address: Department of Chemical Engineering, Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX 78712, USA. 0022-1759/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2007.01.008

42

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

1. Introduction Targeted therapy encompasses a wide variety of different strategies, which can be divided into direct or indirect approaches. Direct approaches target tumorassociated or-specific proteins to alter their signaling either by monoclonal antibodies (mAbs) binding to the relevant antigens or by small-molecule drugs that interfere with these proteins (molecular targeting). Indirect approaches rely on tumor-associated proteins expressed on the cell surface that serve as a target for fusion proteins containing different kinds of effector molecules (Schrama et al., 2006). To deliver a lethal cargo into tumor cells, the targeting antibodies should efficiently internalize into the cells. Similarly, to qualify as targets for such drugs, newly-discovered cell-surface molecules should facilitate the internalization of antibodies that bind to them. Internalization can be studied be several biochemical and microscopy approaches (Casalini et al., 1993; Blake, 2001; Lang et al., 2006). To be an effective drug delivery vehicle, an antibody should not only internalize into the target cells, but should follow an intracellular route ending with the delivery of the drug to the correct intracellular address. Hence, an undisputed proof of the therapeutic potential of such an antibody can be provided by showing that it can, indeed, deliver a cytotoxic drug into such cells, resulting in their death. Such agents are known as immunoconjugates and immunotoxins. (FitzGerald et al., 1988, 2004; Pietersz and McKenzie, 1992; Schrama et al., 2006) To construct reagents with selectivity for certain tumor cells, immunotoxins were initially generated where mAbs or Fab' fragments were chemically linked to potent protein toxins derived from plants or bacteria like ricin, abrin, saporin, Pseudomonas exotoxin (PE), and diphtheria toxin (DT), which combined the selectivity of the carrier moiety with the potency of the toxin moiety. (Pastan and Kreitman, 2002). Immunotoxins kill cancer cells via binding to a surface antigen, internalization and delivery of the toxin moiety to the cell cytosol. In the cytosol, toxins catalytically inhibit a critical cell function and cause cell death. The antibody portion of the immunotoxins targets antigens that are expressed preferentially on the surface of cancer cells. This “first generation” of immunotoxins showed impressive results in vitro but in most cases disappointing anti-tumor effects in animals or humans. The “second generation” of recombinant immunotoxins are antibody-toxin chimeric molecules that are mostly, fully recombinant and consist of a targeting moiety, usually in the form of a single chain antibody, genetically linked to a truncated version of either DT or PE. (Reiter and Pastan, 1998).

Over the years, a large number of antibodies that bind tumor-associated antigens were isolated. Early on, the need for rapid identification of the potential of such an antibody was recognized, since internalization is a pre-requisite for most drug delivery approaches. (Casalini et al., 1993) A proof of internalization can be provided by linking the antibody to a cytotoxic cargo (such as a drug or a toxin) and testing the antibody's ability to deliver its cargo into a target cell. Basically, the first generation of immunotoxins, the antibody–toxin chemical conjugates, could provide such a tool. However, chemical conjugation may not work well with some antibodies, and surely can not be used with polyclonal serum. The generation of a recombinant immunotoxin from each candidate antibody is technically feasible, but extremely laborious. A few “general purpose” agents that could potentially link any IgG to a toxin were reported over the past two decades, most by fusing the IgG Fc-binding protein-A or fragments thereof to various toxins. However, none of these agents proved to be effective in target cell-killing. (Kim and Weaver, 1988; O'Hare et al., 1990; Madshus et al., 1991; Tonevitskii et al., 1991). During the past few years we have been evaluating the therapeutic potential of the anti-Muc1 antibody H23 (Mazor et al., 2005). This study was carried out with the purpose of evaluating the potential of the H23 to deliver a cytotoxic payload to breast cancer cells, with the Fcbinding toxin fusion protein ZZ-PE38 serving as a tool for studying the potential of H23 to deliver a lethal cargo to target cells. The monoclonal anti-ErbB2 antibody FRP5 was chosen for the study as sort of a “positive control” since the potential of the scFv it was derived from to serve as the targeting moiety of recombinant immunotoxins is well established (Harwerth et al., 1992; Schmidt et al., 1996). Both antibodies that are murine IgG1 antibodies were converted to chimeric IgG1 antibodies to facilitate efficient binding to the ZZ domain and a potential for future use for human therapy. For that purpose we constructed a two-plasmid system to clone antibody variable domain for expression as IgG antibodies with the Fc of human IgG1 in transfected mammalian cells. The antibody Fc-binding ZZ domain, a tandem repeated, mutated domain B derived from the antibodybinding protein-A of Staphylococcus aureus had been applied in a multitude of biotechnological applications (Nilsson et al., 1987, 1996). With the aim of evaluating the drug delivery potential of antibodies that bind tumorassociated antigens, we prepared a novel fusion protein, ZZ-PE38 where the ZZ domain is linked to a truncated Pseudomonas exotoxin A. We also describe a twoplasmid system for cloning antibody variable domain for expression as chimeric of human IgG1 antibodies. We further describe the preparation of complexes between

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

ZZ-PE38 and IgGs that bind tumor cells and the specific cytotoxicity of the resultant immunocomplexes. Our results suggest that ZZ-PE38 could prove to be an invaluable tool for the evaluation of the therapeutic potential of antibodies and antibody targets. 2. Materials and methods 2.1. Materials All secondary (HRP-conjugated and fluorescent) antibodies were from Jackson ImmunoResearch Laboratories, USA. Cell culture media and additives were from Beit-Haemek, Israel. Unless stated otherwise, all chemicals, enzyme substrates and MTT reagents were from Sigma, Israel. Nitrocellulose filters were from Schleicher and Schuell, USA. All column chromatography matrices were from Pharmacia, Sweden (Now part of GE healthcare, USA). 2.2. Methods 2.2.1. Construction of mammalian IgH and IgL expression vectors and cloning of immunoglobulin genes to be expressed as a chimeric IgG1 Mammalian vector pMAZ-IgH for human γ1 heavy chain expression and pMAZ-IgL for human κ light

43

chain expression were designed for production of human IgG1 antibodies in mammalian cell culture (Fig. 1). Each vector carries the germline constant domain sequences of the respective heavy or light chain gene including its polyadenylation site located 3′ to the translation termination codon. VH domains are introduced into the IgH expression vector via BssHII and NheI restriction sites, whereas, VL domains are cloned into the IgL expression vector as BssHII and BsiWI restriction fragments. The IgH plasmid carries a neomycin expression cassette for G418 selection, while the IgL plasmid carries a hygromycin B resistance cassette for the isolation of stable transfectants under double drug selection. The strong human cytomegalovirus early promoter controls the expression of both the heavy and light chain genes. In addition, both vectors contain an ampicillin selectable marker and SV40, ColE1 and f1 origin of replication. Our vectors share many features with those described by (Jostock et al., 2004) but in our system the heavy and light chains are carried on separate plasmids, which we find more convenient than a single-all-containing-plasmid in terms of cloning ease. The heavy chain expression vector pMAZ-IgH was constructed on the backbone of pCMV/myc/ER/Neo (Invitrogen, USA). The human gamma 1 constant heavy chain region (CH1–CH3) was recovered by PCR from

Fig. 1. Schematic representation of mammalian expression vectors pMAZ-IgH and pMAZ-IgL. Represented are maps of plasmids pMAZ-IgH for human γ1 heavy chain expression and pMAZ-IgL for human κ light chain expression carrying desired V genes for the production of human IgG1– kappa antibodies in mammalian cell culture.

44

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

Table 1 List of PCR primers Oligonucleotide

Sequence

Hum-CH1-NheI-BACK Hum-CH3-Stop-XbaI-FOR H23-VH-BssHII-BACK H23-VH-NheI-FOR Hum-CK-BsiWI-BACK H23-CL-STOP-XbaI-FOR H23-VK-BssHII-BACK H23-VK-BsiWII-FOR FRP5-VH-BssHII-BACK FRP5-VH-NheI-FOR FRP5-VK-BssHII-BACK FRP5-VK-BsiWI-FOR B1-Nco-BACK Iseq14 ZZ-NcoI-BACK ZZ-HindIII-NotI-FOR

5′-CCACAGGCGCGCACTCCGAGGTCCAACTGCAGGCTAGCACCAAGGGCCCATCGGTC-3′ 5′-TGTGTGTCTAGATTATTTACCCGGGGACAGGG-3′ 5′-CCACAGGCGCGCACTCCGAAGTGAAGCTTGAGGAGTCTGG-3′ 5′-CTTGGTGCTAGCCGAAGAGACAGTGACCAGAGT-3′ 5′-CCACAGGCGCGCACTCCGAAATGGTTCTCACCCGTACGGTGGCTGCACCATCTGTCTTCATCTTC-3′ 5′-TCTCTCTCTAGATTATTAACACTCTCCCCTGTTG-3′ 5′-CCACAGGCGCGCACTCCCAGCTCCAGATGACCCAGTC-3′ 5′-CTTGGTGCTAGCCGAAGAGACAGTGACCAGAGT-3′ 5′-CCACAGGCGCGCACTCCCAGGTACAACTGCAGCAGTCTGG- 3′ 5′-CTTGGTGCTAGCAGAGGAAACGGTGACCGTGGTCC- 3′ 5′-CCACAGGCGCGCACTCCCGACATCCAGCTGCCCAGTC- 3′ 5′-AGCCACCGTACG TTTGATCTCCAATTTTGTCCCCCGAGC- 3′ 5′-CGATGGCCATGGCCGAGGTGCAGCTGGTGGAATCTGG-3′ 5′-CTCAGCTTCCTTTCGGGCTTT-3′ 5′-CCGCTTCCATGG TAGACAACAAATTCAACAAAG-3′ 5′-GTTTAAGCTT TAGCGGCCGCTTTCGGCGCCTGAGCATCATTTAG-3′

Restriction sites are italicized. As common in antibody engineering, BACK primers are 5′ sense primers and FOR primers are 3′ antisense primers.

human lymphoid cDNA, (Azriel-Rosenfeld et al., 2004) using primers Hum-CH1-NheI-BACK and Hum-CH3Stop-XbaI-FOR (all PCR primers are listed in Table 1). After sequence validation, the human constant fragment was inserted via the NheI and XbaI restriction sites in to plasmid pCMV/myc/ER/Neo that had been cut by the same enzymes, resulting in the removal of the myc-tag and the ER retention signal of pCMV/myc/ER/Neo leaving only the signal peptide sequence. Murine H23 VH domain was amplified from plasmid pMALc-NNmuH23(Fv) (Mazor et al., 2005) using primers H23VH-BssHII-BACK and H23-VH-NheI-FOR and introduced into the heavy chain vector as a BssHII/NheI fragment, the resulting vector was named pMAZ-IgHH23. The light chain expression vector pMAZ-IgL was constructed on the backbone of pcDNA3.1/Hygro (Invitrogen, USA). The plasmid DNA sequence between the SspI and XbaI sites was replaced by an SspI/ XbaI DNA fragment recovered from plasmid pCMV/ myc/ER/Neo. The resulting plasmid was named pCMV/ myc/ER/Hygro. The human Kappa constant light chain region was recovered by PCR from human lymphoid cDNA, (Azriel-Rosenfeld et al., 2004) using primers Hum-CK-BsiWI-BACK and Hum-CK-Stop-XbaIFOR. The resulting PCR product was digested with BssHII and XbaI and inserted into the backbone of plasmid pCMV/myc/ER/hygro that was cut with the same enzymes. Murine H23 VK domain was amplified from plasmid pMALc-NN-muH23(Fv) (Mazor et al., 2005) using primers H23-VK-BssHII-BACK and H23VK-BsiWII-FOR and introduced into the light chain

vector as a BssHII/BsiWI fragment, the resulting vector was named pMAZ-IgL-H23. All constructs were validated by DNA sequencing. The plasmids pMAZ-IgH-FRP5 and pMAZ-IgLFRP5 were constructed by replacing the H23 variable domains in the corresponding plasmids with FRP5 variable domains. The anti-ErbB2 FRP5 (Wels et al., 1992) VH and VL domains were obtained by using DNA of plasmid pHEN-FRP5(Fv) (kindly provided by W. Wels) as template in two PCR reactions. The VH region was amplified using primers FRP5-VH-BssHII-BACK and FRP5-VH-NheI-FOR and The VK region was amplified using primers FRP5-VK-BssHII-BACK and FRP5-VK-BsiWI-FOR with the former introducing a BssHII site at the 5′ end and the latter a BsiWI restriction site at the 3′ end. The PCR products were cloned as described above, resulting in plasmids pMAZIgH-FRP5 and pMAZ-IgL-FRP5. In both heavy and light chain plasmids, the BssHII restriction site is located within the ER secretion signal sequence, 5 bp before the open reading frame of the cloned variable domain. In pMAZ-IgH, the NheI site overlaps with the first two codons of the CH1 domain. In pMAZ-IgL, the BsiWI site overlaps with the first two codons of the Ckappa domain. 2.2.2. Transfection of HEK293 cells with pMAZ-IgH and pMAZ-IgL expression vectors and screening of cell culture supernatants of clones expressing chimeric IgGs using dot-blot and ELISA Co-transfections of HEK293 cells with pMAZ-IgH and pMAZ-IgL expression vectors were performed

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

using the non-liposomal transfection reagent FuGene6 (Roche, Belgium) according to the manufacturer's instructions. Briefly, 106 cells were seeded into 6 well plates and 24 h after transfection, about 1000 cells/well were placed into 96-well plates in medium containing 1.2 mg/ml of G418 and 200 μg/ml of hygromycin. Supernatants of single colonies that were grown to near confluence on medium containing selection markers were tested for IgG1 secretion by dot-blot analysis and analyzed in ELISA for antibody binding to MUC1 or to ErbB2. To screen for IgG producing cells, 100 μl supernatants of clones were applied via a vacuum manifold onto a nitrocellulose filter using a dot-blot apparatus. After blocking the membrane with 5% (v/v) non-fat milk in TBS for 1 h at room temperature, the membrane was washed briefly with TBS and incubated with HRP-conjugated goat anti-human antibodies × 10,000 dilution in TBS / 2% milk) for 1 h at room temperature. The membrane was developed using ECL reagents (NEN, USA) according to the supplier's instructions. To determine the amount of chimeric IgG secreted from positive clones, 100 μl supernatants were applied in a two-fold dilution series via a vacuum manifold onto a nitrocellulose filter using a dot-blot apparatus alongside a two-fold dilution series of commercial human IgG standard, starting with a concentration of 1000 ng/ml. After blocking the membrane with 5% (v/v) non-fat milk in TBS for 1 h at room temperature, the membrane was washed briefly with TBS and incubated with HRP-conjugated goat anti-human antibodies diluted × 10,000 in TBS / 2% milk) for 1 h at room temperature. The membrane was developed as described above. IgG1 producing clones were tested for antigen binding in ELISA as described below. 2.2.3. Production and purification of chimeric IgGs from culture media of HEK293 transfected cells Approximately 3 × 106 transfected HEK293 cells were cultured in 75 cm2 flasks containing DMEM, supplemented with 10% fetal calf serum (FCS), 1.2 mg/ ml G418 and 200 μg/ml hygromycin, at 37 °C, 5% CO2, in a humidified incubator. The culture was allowed to grow to 80% confluence followed by a gradual starvation of the cells to FCS, in a two-fold dilution series for a period of 24 h of each dilution. The cells were deprived of FCS 72 h prior to harvesting. Chimeric IgG was purified using protein-A sepharose (Amersham Biosciences, Sweden) chromatography as recommended by the supplier. Protein-containing fractions were combined, dialyzed against 5 l of Phosphatebuffered saline (PBS) (16 h, 4 °C), sterile filtered and

45

stored at 4 °C. For immunoblotting, purified IgGs were separated by 12%/SDS polyacrylamide gel electrophoresis under reducing conditions and electro-transferred onto nitrocellulose membrane. Chimeric IgGs were detected with HRP-conjugated goat anti-human antibodies. The membrane was developed using ECL reagents (NEN, USA) according to the supplier's instructions. 2.2.4. Evaluation of binding by ELISA and whole-cell ELISA Analysis of MUC1 binding by ELISA was carried out as follows, ELISA plates were coated with a 10-fold dilution of MUC1 transfected DA3 cells conditioned medium (Baruch et al., 1999) diluted in 50 mM NaHCO3 buffer (pH 9.6) at 4 °C for 20 h and blocked with 2% (v/v) non-fat milk in PBS for 2 h at 37 °C. All subsequent steps were done at room temperature. 2 μg/ ml of purified chimeric H23 IgG1 and murine H23 mAb (Tsarfati et al., 1989) were applied onto the plates in a two-fold dilution series and tested for their affinity to MUC1. Following incubation the plates were washed three times with PBST. HRP-conjugated goat antihuman and HRP-conjugated goat anti-mouse antibodies were used as secondary antibodies diluted × 10,000 in PBST. The ELISA was developed using the chromogenic HRP substrate TMB and color development was terminated with 1 M H2SO4. The results were plotted as absorbance at 450 nm and the binding-affinity was estimated as the IgG concentration that generates 50% of the maximal signal. Cellular MUC1 binding by chH23 IgG1 and murine H23 mAb, or cellular ErbB2 binding by chFRP5 IgG1, were tested by whole-cell ELISA as follows; cell lines used were the human breast carcinoma T47D for MUC1 and SKBR3 for ErbB2. Following trypsinization, cells were washed once with 2% fetal calf serum, 0.05% NaN3 in PBS. Approximately 106 cells were divided into individual immunotubes (Nunc, Sweden). After washing twice with PBS, HRP-conjugated goat antihuman and HRP-conjugated goat anti-mouse (× 2000 dilution) were added to the appropriate tubes for 1 h at 4 °C. Detection of cell bound antibodies was performed by addition of 0.5 ml of the chromogenic HRP substrate TMB to each tube and color development was terminated with 0.25 ml of 1 M H2SO4. Finally, the tubes were centrifuged for 10 min at 4000 rpm and color intensity of supernatants was measured at 450 nm. The binding-affinity was estimated as the IgG concentration that generates 50% of the maximal signal. In the case of chH23 IgG1, to confirm specificity, antibodies were incubated with an excess of MUC1 protein from DA3-

46

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

MUC1-transected cell line as competitor for 1 h at room temperature prior to incubation with the cells and then were added to the cell tubes for 1.5 h at 4 °C. 2.2.5. Flow cytometry Cellular MUC1 binding by chH23 IgG1 and murine H23 mAb was tested by flow cytometry. Cell lines used were the mouse cell line DA3, the MUC1-transfected cell line DA3-MUC1, (Baruch et al., 1999) the breast carcinoma lines T47D and MCF7 and the human kidney cell line HEK293 that served as negative control. Approximately 5 × 105 cells were used in each experiment. After trypsinization, cells were washed twice in 2% fetal calf serum, 0.05% NaN3 in PBS (FACS buffer). After washing twice with FACS buffer, FITC-labeled goat anti-human and FITC-labeled goat anti-mouse (× 50 dilution) was added to the appropriate tubes for 45 min at 4 °C. Detection of bound antibodies was performed by means of flow cytometry on a FACSCalibur (Becton Dickinson, CA) and results were analyzed with the CELLQuest program (Becton Dickinson). In the case of MUC1, to confirm specificity, antibodies (10 μg/ml) were incubated with or without an excess of MUC1 protein from DA3-MUC1-transected cell line, for 1 h at room temperature prior to incubation with the cells and then were added to the cell tubes for 1 h at 4 °C. 2.2.6. Analysis of internalization using confocal microscopy Internalization of chH23 IgG1 and chFRP5 IgGs was studied using confocal microscopy. The human breast carcinoma T47D (for chH23) or SKBR3 (for chFRP5) cell lines were used to evaluate the internalization capabilities of the antibodies at 4 °C and 37 °C as follows; sterile 24 mm cover slips were placed in a 6 well plate and incubated for 1 h at RT with 800 μl of 10 μg/ml poly-L-lysine followed by two washes with 1 ml PBS. Approximately 4 × 105 cells were seeded on the cover slips in each well and cells were grown to 40%–50% confluence in DMEM supplemented with 10% FCS. Cells tested for internalization at 4 °C were pre-incubated with medium supplemented with 0.5% NaN3 for 2 h in 4 °C prior to the addition of the antibody (5 μg/ml) and further incubation of 1 h in 4 °C. Cells tested for internalization at 37 °C were incubated with 5 μg/ml of the antibody in medium for 1 h at 37 °C in a humidified atmosphere of 95% air and 5% CO2. Following the incubation with the antibody, the cells were gently washed twice with PBS to remove excess antibody and incubated for 2 h under the same conditions of the previous step. Next, the cells were

gently washed twice with PBS and fixed in two sequential steps of incubation for 10 min with ice cold methanol followed by 10 min incubation with ice cold acetone. Following the fixation step, the cells were gently washed with PBS and blocked with 10% normal goat serum in PBS for 25 min at RT. The blocking solution was aspirated and FITC-labeled goat antihuman antibodies (× 150 dilution) were added and cells were further incubated for 2 h at RT. Finally, the cells were gently washed three times with PBS and staining pattern (membranous or intracellular) images were acquired using a LSM 510 laser scanning confocal microscope (Vontz 3403B). 2.2.7. Preparation of the ZZ-PE38 fusion protein Plasmid pET22-NN-ZZ-PE38 was designed for the expression of soluble ZZ-Pseudomonas exotoxin A (PE38) fusion protein secreted to the periplasm of E. coli cells. Initially, a DNA fragment coding for a the recombinant immunotoxin B1(Fv)-PE38 (Benhar and Pastan, 1995) was amplified by PCR using the plasmid pB1(Fv)-PE38 (Benhar and Pastan, 1995) as template with primers Iseq14 and B1-NcoI-BACK (all PCR primers are listed in Table 1). The PCR product was digested by NcoI and EcoRI and inserted into the plasmid backbone of pET22b (Novagen, USA) that had been cut by the same enzymes. This resulted in plasmid pET22b-B1(Fv)-PE38, in which the immunotoxin coding DNA is preceded by the plasmid-encoded pelB secretion signal sequence. Next, the B1 scFv fragment was removed by digesting pET22b-B1(Fv)-PE38 by NcoI and HindIII and replaced by a DNA fragment carrying the ZZ domain that was obtained by PCR amplification of plasmid pCANTAB-ZZ (Yacoby et al., 2006) as template with primers ZZ-NcoI-BACK and ZZ-HindIII-NotI-FOR followed by digestion by the same enzymes. This resulted in plasmid pET22-NN-ZZPE38 that carries an in-frame fusion of ZZ to PE38. For expression, E. coli BL21 (DE3) cells transformed with plasmid pET22-NN-ZZ-PE38 were grown in 1 l of SB medium (12 g Bacto-Tryptone, 24 g Yeast extract, 6.3 g Glycerol, 12.5 g K2HPO4, 3.8 g KH2PO4 per 1 l water) supplemented with 100 μg/ml ampicillin, 0.5% (w/v) glucose and 0.4 g/l MgSO4 at 37 °C. When the cells reached A600 nm of 2.5 they were induced for protein overexpression with 1 mM IPTG at 30 °C for 3 h. Following induction, the cells were collected by centrifugation (15 min, 4000 rpm at 4 °C) and the periplasmic fractions were prepared by gently resuspending the cell pellet using glass beads in 200 ml of ice-cold 20% sucrose, 30 mM Tris–HCl (pH 7.4),

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

1 mM EDTA and left on ice for 15 min. Next, cells were collected by centrifugation (15 min, 6000 rpm at 4 °C), the supernatant was discarded off and the pellet was gently re-suspended in 200 ml of ice cold sterile water and left on ice for 15 min. Following incubation on ice, the periplasmic fraction was obtained by centrifugation of the spheroplasts (15 min, 7000 rpm at 4 °C). The resulting periplasmic fraction was adjusted to 20 mM Tris–HCl (pH 7.4), 1 mM EDTA and the ZZ-PE38 fusion protein was purified in two sequential chromatography steps of Q-Sepharose and mono-Q anion exchange columns using fast protein liquid chromatography (FPLC). A Q-sepharose column (∼ 8 ml resin per 200 ml of periplasm extract) was prepared and used for anion exchange chromatography according to the supplier's recommendations using FPLC. Buffer A containing: 20 mM Tris–HCl pH 7.4, 1 mM EDTA, was used for equilibration and washing. Buffer B containing: 1 M NaCl in buffer A, was used as an elution buffer. Loading, washes and elution flow rate were 5 ml/min. After loading the protein, the column was washed with 10% buffer B (0.1 M NaCl in buffer A). The protein was eluted with 30% buffer B (0.3 M NaCl in buffer A), 10 ml fraction were collected. Protein concentration was determined in each fraction. The protein containing fractions were analyzed by SDS/PAGE. The ZZ-PE38containing fractions were pooled (Q-sepharose pool) and diluted with 12.5 volumes of buffer A before applying on to a monoQ column. A 1 ml monoQ column was used for further purification of the ZZ-PE38 fusion protein using FPLC. Loading, washes and elution flow rate were at 2 ml/min. After loading the protein, the column was washed with 10 ml of buffer A. The column was developed with a linear gradient of 0–100% buffer B over 60 ml. 2 ml fractions were collected. Protein concentration was determined, SDS gel analysis was performed, and relatively pure and concentrated protein fractions were pooled (monoQ pool). The pooled fractions were concentrated using Amicon Centrifugal Filter Devices (30 kDa cutoff, Millipore, USA) to a final concentration of 5 mg/ml, sterile filtered and stored at 4 °C. 2.2.8. Preparation of IgG-ZZ-PE38 immunocomplex The immunocomplex of chH23 IgG1, chFRP5 IgG1 and commercial, protein-A purified human IgG (used as control) antibodies with ZZ-PE38 fusion protein was performed as follows; 0.5 ml of 4.5 mg/ml IgG in PBS were mixed with 0.5 ml of 4.5 mg/ml ZZ-PE38 fusion protein in PBS (3 fold molar excess of ZZ-PE38 over

47

Fig. 2. Analysis of HEK293 cell clones expressing chimeric H23 IgG and evaluation of antigen binding. A. Dot-blot analysis screen of spent culture medium of clones expressing chimeric IgG. 100 μl supernatants of individual clones were applied in duplicate onto a nitrocellulose filter using a dotblot apparatus. ChH23 IgG1 antibodies were detected using HRP-conjugated goat anti-human IgG1 as secondary antibodies. The membrane was developed using ECL reagents and exposure to X-ray film. B. Immunoblot of spent culture medium of positive clones for determination of heavy and light chain production level, lane 1–4 candidate clones, lane 5 commercial human IgG at 0.5 μg/ml. C. Analysis of MUC1 binding by individual clones by ELISA. Microtiter plates were coated with MUC1-containing spent culture medium and MUC1 binding assays were performed by applying 100 μl of spent culture medium of positive clones onto the plates. HRP-conjugated goat anti-human IgG was used as secondary antibodies. The ELISA was developed using the chromogenic HRP substrate TMB. The error bars represent standard deviations of three independent experiments. D. Determination of chH23 secretion levels. 100 μl supernatants of clone B2 were applied in a two-fold dilution series onto a nitrocellulose filter using dot-blot alongside a two-fold dilution series of commercial human IgG standard starting with a concentration of 1000 ng/ml. HRPconjugated goat anti-human IgG was used as secondary antibodies.

48

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

IgG) in an Eppendorf tube that was rotated hear-overhead for 16 h at 4 °C. Separation of excess ZZ-PE38 from the IgG-ZZ-PE38 complex was performed by applying the sample onto a 25 ml Superdex 75 size exclusion column (GE Healthcare, USA). The column was initially equilibrated with 30 ml of PBS followed by loading of 1 ml of 4.5 mg/ml total protein into the sample loop of the FPLC system. 30 ml PBS was run at a flow rate of 1 ml/min and 1 ml fractions were collected. According to protein size marker calibration, fractions 9–10 contained the monomeric chH23-ZZ-

PE38 immunocomplex, while fractions 11–12 contained the excess free ZZ-PE38 protein. The protein concentration and purification at the relevant fractions were determined as described above. The immunocomplex-containing fractions were concentrated using Centrifugal Filter Devices (30 kDa cutoff) to a final concentration of 3 mg/ml, sterile filtered and stored at 4 °C. The resulting immunocomplexes were chH23-ZZPE38, chFRP5-ZZ-PE38 and the control hIgG-ZZPE38. 2.2.9. Binding analysis of IgG-ZZ-PE38 immunocomplexes by ELISA and flow cytometry Evaluation of cellular MUC1 binding activity of the chH23-ZZ-PE38 immunocomplex and of the chFRP5ZZ-PE38 immunocomplex by whole-cell ELISA was performed essentially as described above using the same cell lines. Anti-PE sera (× 200 dilution; Ira Pastan, personal communication) mixed with HRP-conjugated goat anti-rabbit antibodies (× 1500 dilution) were used for the detection of bound immunocomplex. Evaluation of cellular MUC1 binding by the chH23ZZ-PE38 immunocomplex was tested by flow cytometry essentially as described above. The cell line used was the human breast carcinoma T47D. To confirm specificity, chH23-ZZ-PE38 immunocomplex was incubated in the presence or absence of 10-fold excess of un-conjugated

Fig. 3. Purification of chimeric IgGs and evaluation of antigen binding. A. Purification of soluble chH23 IgG1 from culture media of HEK293 clone B2 cells by protein-A column chromatography, lane 1, total culture media, lane 2, protein-A flow-through, lane 3, purified chH23 IgG1 clone B2 antibody at a yield of 20 mg/l of culture media. Proteins were separated on a 12%/SDS polyacrylamide gel under reducing conditions and visualized by staining with GelCode Blue®. B. Immunoblot analysis of purified chH23 IgG1 clone B2 antibody using HRP-conjugated goat antihuman IgG. The arrows mark the size and position of human IgG1 heavy and light chains. C. Comparative analysis of MUC1 binding-affinity by chH23 IgG1 and murine H23 mAb by ELISA. Microtiter plates were coated with MUC1-containing spent culture medium. MUC1 binding assays were performed with 2 μg/ml of purified antibodies (H23 mAb, filled triangles, chH23 IgG1, filled squares) in a two-fold dilution series. HRP-conjugated goat anti-human and HRP-conjugated goat anti-mouse antibodies were used as secondary antibodies. D. Comparative cellular MUC1 binding-affinity by chH23 IgG1 (filled squares) and murine H23 mAb (filled triangles) tested by whole-cell ELISA. The MUC1 expressing cells used were the human breast carcinoma T47D cell line. To confirm specificity, antibodies were incubated in the presence of MUC1 protein prior to incubation with the cells (chH23 IgG +competitor, open squares; H23 mAb + competitor, open triangles). The binding-affinity was estimated as the IgG concentration that generates 50% of the maximal signal. E. Cellular ErbB2 binding by chFRP5 IgG1 (triangles) and Herceptin® mAb (squares) was evaluated on the human breast carcinoma SKBR3 cell line by whole-cell ELISA. The error bars represent standard deviations of three independent experiments.

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

chH23 IgG1 prior to incubation with the cells. Rabbit anti-PE sera mixed with FITC-conjugated goat antirabbit antibodies (× 50 dilution) were used for the detection of bound chH23-ZZ-PE38 immunocomplex.

49

(A570 of treated sample / A570 of untreated sample) × 100. The IC50 values were defined as the immunotoxin concentrations inhibiting cell growth by 50%. 3. Results

2.2.10. Cell-viability assay The in vitro cell-killing activity of chH23-ZZ-PE38 and of chFRP5-ZZ-PE38 immunocomplexes was measured by an MTT assay. Tested cells were seeded in 96well plates at a density of 1 × 104 cells/well in DMEM supplemented with 10% FCS. Various concentrations of immunocomplexes or of control proteins were added to triplicate samples and the cells were incubated for 48 h at 37 °C in 5% CO2 atmosphere. After 48 h, the media was replaced by immunocomplex-free media (125 μl per well) containing 5 mg/ml MTT (Thiazolyl Blue Tetrazoliam Bromide, dissolved in PBS) reagent and the cells were incubated for another 4 h. MTT-formazan crystals were dissolved by the addition of 20% SDS, 50% DMF, pH 4.7 (100 μl per well) and incubation for 16 h at 37 °C in 5% CO2 atmosphere. Absorbance at 570 nm was recorded on an automated microtiter plate reader. It was established that optical density was directly proportional to the cell number up to the density reached by the end of the assay. Identical concentrations and combinations were tested in three separate wells per assay and the assay was performed at least three times. The results were expressed as percentage of living cells relatively to the untreated controls that were processed simultaneously using the following equation:

3.1. Preparation of chimeric H23 and FRP1 IgG derivatives The mammalian expression vectors for human or chimeric IgG1 derivatives, pMAZ-IgH for human γ1 heavy chain expression and pMAZ-IgL for human κ light chain expression are described in Fig. 1. After cotransfection of HEK293 cells with the heavy and light chain gene-containing plasmids, cells were grown in selective media in 96-well plates. Approximately 40% of the wells showed cell growth after 10 days in culture. Supernatants of clones growing on medium containing selection drugs were tested for IgG secretion by Dot-Blot (Fig. 2A) and immunoblot (Fig. 2B). Approximately 90% of the wells with selected cells were positive for IgG production, with secretion levels of chimeric IgG ranging between 0.1 and 20 μg/ml spent culture medium. IgG producing clones were analyzed in ELISA for antibody binding to MUC1 (Fig. 2C), showing a significant correlation between antibody secretion level and the intensity of MUC1 binding. Positive clone B2 that secreted at 20 μg/ml (Fig. 2D) was selected for further evaluation as chimeric H23 IgG1 (chH23).

Fig. 4. Binding of chH23 IgG1 to cellular MUC1. A. Flow-cytometry analysis of cellular MUC1 binding. Comparison between H23 mAb ( panel a.) and chH23 IgG1 ( panel b.) for their binding to cellular MUC1. To confirm specificity, antibodies (10 μg/ml) were incubated with or without an excess of MUC1 protein prior to incubation with the cells. FITC-labeled goat anti-human and FITC-labeled goat anti-mouse were used as secondary antibodies. Filled areas, negative control (secondary antibody alone); black line, binding of specific antibody; grey line, competition for cell binding with soluble MUC1 protein.

50

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

Chimeric H23 IgG1 antibody was purified from cell line B2 grown in FCS starvation media as described in Materials and methods. We found that stepwise starvation of the cells to FCS not only minimized the contamination with bovine IgG upon purification on protein-A column, but also increased the amount of antibody secreted to the medium by the FCS deprived cells (data not shown). Under these conditions, more than 95% pure chimeric H23 IgG1 protein was obtained based upon separation on 12%/SDS-PAGE under reducing conditions (Fig. 3A) and verification of human IgG heavy and light chain production by immunoblot (Fig. 3B). Chimeric anti-

MUC1 H23 antibody was purified at a total yield of 20 mg/l of spent culture medium. The anti-ErbB2 antibody FRP5 is well known for its potential to internalize into ErbB2-expressing cells as it was extensively studied as the targeting moiety of recombinant immunotoxins (Harwerth et al., 1992; Wels et al., 1992). The anti-ErbB2 FRP5 VH and VL domain were cloned into pMAZ-IgH and pMAZ-IgL expression vectors as described in Materials and methods. The resulting plasmids were named pMAZ-IgH-FRP5 and pMAZ-IgLFRP5 respectively. Transfection, production and purification of chFRP5 IgG1 from mammalian HEK293 cell line

Fig. 5. Analysis of antibody internalization by confocal microscopy. Internalization studies of chH23 IgG1 in human breast carcinoma T47D cells and of chFRP5 IgG1 into SKBR3 cells was evaluated at 4 °C (A and C, respectively) and 37 °C (B and D, respectively) using confocal microscopy as described in Materials and methods. Serial cuts of the analyzed cells are presented.

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

were performed essentially as described above for chH23. Positive clone G1 that secreted at 25 μg/ml was selected for further evaluation of this chimeric anti-ErbB2 IgG1. 3.2. Analysis of antigen binding by chH23 and chFRP5 IgG1 derivatives To compare the apparent binding affinities of chH23 IgG1 with that of murine H23 mAb, we performed a comparative half-maximal binding assay using ELISA. The results (Fig. 3C) suggest that their apparent MUC1 binding affinities are comparable, as the apparent binding-affinity of the two antibodies were 0.2 nM for the murine mAb and 0.3 nM for chH23 IgG1. To further compare their binding affinities we tested cellular MUC1 binding by whole-cell ELISA. As with the ELISA results, the results of this assay showed that both antibodies bind the MUC1 positive tumor cells with a similar apparent affinity, 0.25 nM for the murine mAb and 0.3 nM for chH23 (Fig. 3D), which are in agreement with the binding data to MUC1 protein. Furthermore, we demonstrated the specificity of the two antibodies to cellular MUC1, as binding of the MUC1-specific antibodies to the cells could be competed off with an excess of soluble MUC1 protein (Fig. 3D). The binding capabilities of the chimeric and murine antibodies were further evaluated by flow cytometry. Our results show that chH23 (Fig. 4A panel a) stains the breast cancer cell lines and the MUC1 transfected cells in the same pattern as the murine H23 mAb (Fig. 4A panel b) whereas no binding was seen with the control DA3 and HEK293 cell lines (negative controls). We further demonstrated the specificity of the two antibodies to cellular MUC1, as binding of the both antibodies to MUC1 positive cells was

51

competed off following incubation of the antibodies with an excess of MUC1 protein prior to incubation with the cells. When evaluated for specific cellular-ErbB2 binding, chFRP5 IgG1 exhibited similar binding characteristics as the commercial FDA-approved humanized anti-ErbB2 mAb Herceptin® (Trastuzumab) (Fig 3E). Their binding sites, however, do not overlap, and FRP5 has no affect on ErbB2 expressing tumor cells as an unarmed IgG. 3.3. Evaluation of antibody internalization using confocal microscopy Internalization of chH23 into MUC1-expressing T47D cells was evaluated at 4 °C (condition that prevent internalization) and at 37 °C (condition permissive for internalization). The confocal microscopy results are presented by serial cuts of the analyzed cells with the purpose of distinguishing between the membrane fluorescence with fluorescence within the cell. Internalization is typically characterized by bright fluorescent vesicles within the cell cytoplasm in the mid-sections together with a decrease in membranous fluorescence. Our results (Fig. 5A and B) showed that no detectable internalization of chH23 IgG1 could be observed. Under these experimental conditions, the antibody remained bound to the cell surface so only membranous fluorescence could be observed. To confirm that chFRP5 not only binds specifically to ErbB2-expressing cells but also internalizes into the cells following receptor binding we analyzed its capability to internalize into the highly ErbB2-expressing SKBR3 cells using confocal microscopy. The internalization of chFRP5 to the SKBR3 cells was evaluated at 4 °C and at 37 °C as described above for chH23. The confocal microscopy

Fig. 6. Purification of soluble ZZ-PE38 fusion protein by anion exchange chromatography and of chH23-ZZ-PE38 and chFRP5-ZZ-PE38 immunocomplexes by size-exclusion chromatography. A. Periplasmic purification of ZZ-PE38 fusion protein was obtained by two sequential chromatography steps of Q-Sepharose and mono-Q anion exchange columns using FPLC. MW, molecular weight marker; lane 1, soluble proteins from the induced cells extract, 15 μg of ZZ-PE38 periplasm extract; lane 2, purified ZZ-PE38, 5 μg of protein. B. Purification of chH23-ZZ-PE38 immunocomplex from excess ZZ-PE38 was obtained by gel filtration of the mixed sample using FPLC. MW, molecular mass marker; lane 1, purified, 2 μg of chH23-ZZ-PE38 immunocomplex. C. Same as B for chFRP5-ZZ-PE38, but 4 μg were loaded in lane 1. Proteins were separated on 12%/SDS polyacrylamide gels under reducing conditions and visualized by staining with GelCode Blue®. The arrows mark the positions of ZZ-P38 and chimeric IgG heavy and light chains.

52

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

results (Fig. 5C and D) clearly show that at 37 °C chFRP5 had internalized into the ErbB2-expressing cells as indicated by the appearance of bright fluorescence vesicles within the cell cytoplasm in the mid-sections, while, no detectable internalization could be observed at 4 °C.

3.4. Production of ZZ-PE38 fusion protein and preparation of chH23-ZZ-PE38 immunocomplexes An expression vector for bacterial production of ZZPE38 was constructed and soluble ZZ-PE38 fusion from

Fig. 7. Analysis of cellular MUC1 binding by chH23-ZZ-PE38. Cellular MUC1 binding was evaluated on human breast carcinoma T47D cell line. A. Cells were incubated with serial dilutions of chH23-ZZ-PE38 (circles) or chH23 IgG1 (triangles) for 1.5 h at 4 °C. Rabbit anti-PE sera mixed with HRP-conjugated goat anti-rabbit antibodies were used for the detection of bound chH23-ZZ-PE38 immunocomplex, while HRP-labeled goat antihuman antibodies were used for the detection of bound chH23 IgG1. The error bars represent standard deviations of three independent experiments. B. Flow-cytometry analysis of chH23-ZZ-PE38 immunocomplex. Cells were incubated with 10 μg/ml of chH23-ZZ-PE38 or control hIgG-ZZ-PE38 for 1.5 h at 4 °C. To confirm specificity, chH23-ZZ-PE38 immunoconjugate was incubated in the presence or absence of 10-fold excess of un-conjugated chH23 IgG1 prior to incubation with the cells. Rabbit anti-PE sera mixed with FITC-labeled goat anti-rabbit antibodies were used for the detection of bound chH23-ZZ-PE38 immunotoxin. Filled areas, negative control; black line, binding of specific antibody; grey line, competition for binding with chH23 IgG1. C. Cellular ErbB2 binding-affinity by chFRP5-ZZ-PE38 immunocomplex (circles) tested by whole-cell ELISA. Cells were processed and the results are presented as in A. The binding-affinity was estimated as the IgG or immunocomplex concentration that generates 50% of the maximal signal.

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

53

MUC1 positive tumor cells with comparable affinity to the un-conjugated chH23 IgG1 (Fig. 7A). We further evaluated the binding of chH23-ZZ-PE38 immunocomplex to MUC1 expressing cells by flow cytometry. Our results (Fig. 7B) show that chH23-ZZPE38 stains the MUC1 positive tumor cells while the control human IgG-ZZ-PE38 immunocomplex that was prepared from a commercial human IgG pool, and purified in the same way showed no binding to the tumor cells. Furthermore, we demonstrated the specificity of chH23-ZZ-PE38 immunocomplex to cellular MUC1, as binding of the anti-MUC1 immunotoxin to the tumor cells could be competed off with the addition of 10-fold excess of un-conjugated chH23 IgG1. The chFRP5-ZZ-PE38 immunocomplex was prepared similarly and analyzed by 12%/SDS-PAGE under reducing conditions (Fig. 6C). The binding of the chFRP5-ZZPE38 immunocomplex to ErbB2-expressing cells was evaluated by whole-cell ELISA. The results (Fig. 7C) show that the chFRP5-ZZ-PE38 immunocomplex retained its binding activity to cellular ErbB2-expressing SKBR3 cells with similar apparent affinity as the unconjugated form of chFRP5 IgG1 (compare to 3E). Fig. 8. Specific cytotoxicity of chH23-ZZ-PE38 and control proteins for MUC1 cell lines. T47D (A) and MCF7 (B) tumor cells were incubated for 48 h with serial dilutions of chH23-ZZ-PE38 (circles), hIgG-ZZ-PE38 (squares) or ZZ-PE38 (triangles). The relative number of viable cells was determined using an enzymatic MTT assay. Each point represents the mean of a set of data determined in triplicate in three independent experiments. The error bars represent standard deviations of three independent experiments.

periplasmic fractions was purified (Fig. 6A). A complex of chimeric IgG1 antibodies with the ZZ-PE38 fusion protein was prepared as described in Materials and methods. Excess ZZ-PE38 was separated from the chH23-ZZ-PE38 complex by size-exclusion chromatography. The purified chH23-ZZ-PE38 immunocomplex was analyzed by 12%/ SDS-PAGE under reducing conditions (Fig. 6B). This fraction contained the constituents of the immunocomplex at seemingly equimolar amounts. The excess ZZ-PE38 eluted from the column in later fractions (not shown). 3.5. Binding properties of chH23-ZZ-PE38 and chFRP5-ZZ-PE38 immunocomplexes We evaluated the binding of chH23-ZZ-PE38 immunocomplex to cells by whole-cell ELISA. The results of this assay showed that the complex of the ZZ-PE38 fusion protein with chH23 was stable, furthermore, the conjugation process did not undermine chH23 antigen recognition as the immunocomplex retained its binding capabilities to

3.6. Cytotoxic activity of chH23-ZZ-PE38 and chFRP5ZZ-PE38 immunocomplexes To evaluate the cytotoxic activity of chH23-ZZ-PE38 immunocomplex towards MUC1-expressing cells, we performed in vitro cell-killing experiments. T47D and MCF7 human breast carcinoma cell lines were incubated for 48 h with various concentrations of chH23-ZZ-PE38 or with the control proteins. The relative number of viable cells in comparison with cells grown in the absence of toxin was determined using an enzymatic MTT assay. The results Table 2 In vitro cell-killing activity and receptor density on ErbB2-expressing cells Cell line

Number ChFRP5- IC50 (ng/ml) ZZ-PE38 ChFRP5 of ErbB2 ZZ-PE38 hIgG-ZZreceptors ⁎ PE38

SKBR3 A431 T47D MCF7 MDAMB231

1.5 × 106 2 × 104 ∼ 104 b1000 b1000

3.5 1.8 1000 600 N10,000

⁎ source data: SKBR3 (Hynes et al., 1989). A431 (Schmidt et al., 1996). T47D (Schmidt et al., 1996). MCF7 (Benz et al., 1989). MDA-MB231 (Wels et al., 1992).

N10,000 3500 N10,000 N10,000 N10,000

N10,000 N10,000 N10,000 N10,000 N10,000

N10,000 N10,000 N10,000 N10,000 N10,000

54

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

(Fig. 8) show that both the highly MUC1-expressing T47D cells and the moderate MUC1-expressing MCF7 cells showed almost no sensitivity to the chH23-ZZ-PE38 immunocomplex with an IC50 of ∼5 μg/ml for T47D cells and no detectable sensitivity for MCF7 cells with the concentrations of chH23-ZZ-PE38 that were used in this assay. Furthermore, comparable results were obtained with the control hIgG-ZZ-PE38 immunocomplex that was shown above as negative for MUC1 binding.

To evaluate the cytotoxic activity of chFRP5-ZZ-PE38 immunocomplex, we performed cell-killing experiments on several human tumor cell lines expressing varying levels of ErbB2 receptors as shown in Table 2. Cells were incubated for 48 h with varying concentrations of chFRP5-ZZ-PE38 and the corresponding control proteins and the relative number of viable cells was determined using an MTT assay. The results are shown in Fig. 9 and the IC50 values are also summarized in Table 2. The

Fig. 9. Specific cytotoxicity of chH23-ZZ-PE38 and control proteins for ErbB2 cell lines. SKBR3 (A) A431 (B) T47D (C) MCF7 (D) and MDAMB231 (E) tumor cells were incubated for 48 h with serial dilutions of chFRP5-ZZ-PE38 (filled circles), hIgG-ZZ-PE38 (filled squares), chFRP5 IgG1 (open circles) or ZZ-PE38 (open triangles). The relative number of viable cells was determined using an enzymatic MTT assay. Each point represents the mean of a set of data determined in triplicate in three independent experiments. The error bars represent standard deviations of three independent experiments.

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

results indicated that chFRP5-ZZ-P38 was cytotoxic for all five cell lines tested with IC50 values that in most cases correlated with the levels of ErbB2 expression among the different tumor cells. SKBR3 and A431 tumor cells were most sensitive to the immunoconjugate with IC50 values of 3.5 ng/ml and 1.8 ng/ml respectively. T47D and MCF7 cells were only moderately sensitive to chFRP5-ZZ-PE38 while MDA-MB231 cells showed very low sensitivity which in some reflects their lower level of ErbB2 expression of approximately 103 or less receptors per cell. Taken together, the cell-killing activities achieved with chFRP5-ZZ-PE38 are a consequence of specific targeting and internalization of the immunocomplex into the tumor cells as none of the separable components alone was significantly toxic towards any of the tumor cells as well as the negative control hIgG-ZZ-PE38. We further demonstrated the specificity of the immunocomplex in targeting and killing ErbB2-expressing tumor cells, since the cytotoxic activity could be competed off in the presence of excess chFRP5 IgG1. 4. Discussion This study was carried out with the purpose of evaluating the potential of the anti-MUC1 antibody H23 to deliver a cytotoxic payload to breast cancer cells, and to prepare a chimeric IgG1 derivative of H23 that will be suitable for further development for human therapy. In the course of the study we prepared the Fc-binding toxin fusion protein ZZ-PE38 as a tool for studying the potential of antibodies to deliver a lethal cargo to target cells. FRP5 was chosen for the study as sort of a “positive control” since the potential of the scFv it was derived from to serve as the targeting moiety of recombinant immunotoxins is well established (Harwerth et al., 1992; Schmidt et al., 1996). Our earlier study suggested the monovalent scFvs derived from H23, whether murine or humanized are lower ×100 in binding avidity in comparison to the H23 mAb (Mazor et al., 2005). Furthermore, recombinant immunotoxins we prepared from H23 murine and humanized scFvs failed to kill MUC1-expressing cells (unpublished results). As a result, and since murine IgGs are not favorable candidates for clinical development (Hwang and Foote, 2005), we decided to evaluate H23 in the form of a chimeric IgG. We prepared a chimeric IgG1 derivative of H23 by the design of two mammalian expression vectors for the expression of the murine VH and VK domains linked to human γ1 and κ constant domains, respectively. For stable transfection of mammalian vectors pMAZ-IgH and pMAZ-IgL we chose the human kidney HEK293 cell line. This cell line was

55

selected for its outstanding capabilities in both accepting foreign DNA and the production of recombinant proteins (Pham et al., 2003). An additional benefit of this cell line is its rapid growth rate that enables the construction of an IgG producing cell line in less than a month (our unpublished results). Several mammalian cell lines have been successfully used for stably producing humanized or chimeric antibodies, include the Chinese hamster ovary cell line CHO, and the mouse myeloma cell lines Sp2/0 and NS0 (Bebbington et al., 1992; Trill et al., 1995; Sanna, 2002; Yoo et al., 2002). However, stable production of recombinant IgG in HEK293 cells was yet to be done. The selection of HEK293 cell for production of chimeric H23 (chH23) proved to be very effective as approximately 40% of the cells were stably transfected with both mammalian vectors and of these, approximately 90% were positive for IgG production, with secretion levels of chimeric IgG between 0.1 and 20 μg/ml. The expression levels we obtained for chH23 and for chFRP5 exceed those reported for stably IgG producing COS cells with expression levels of 3 μg/ml (Ames et al., 1995) and CHO cells expressing at 0.75 μg/ml (Henderikx et al., 2002) using commercial mammalian expression vectors. Furthermore, our overall antibody expression levels are high or in case comparable to those obtained with transiently transfected cells (Durocher et al., 2002; Jostock et al., 2004; Tsurushita et al., 2005). Expression levels of stable transfectants producing up to 50 μg/ml humanized antibodies was reported only for mouse myeloma cell lines Sp2/0 and NS0 (Yoo et al., 2002) that were grown in exhausted serum-free medium. Nevertheless, the production level of humanized antibodies can ultimately reach more than 1 mg/ml in bioreactors applying fed-batch process and media optimization (Xie and Wang, 1997; Sauer et al., 2000). On the other hand, (Simmons et al., 2002) recently described a method for high level expression of fulllength IgG in E. coli, which yielded in vivo plasma halflives comparable to antibodies derived from mammalian expression systems. These antibodies were not glycosylated, and did not bind to complement factor 1q (C1q) or Fc-gamma receptor I (FcγRI), thus making them incapable of mediating effector functions. This, however, will not limit the potential of bacterially-expressed IgGs to be complexed with ZZ-PE38. When evaluated for MUC1 binding, chH23 IgG1 exhibited the same binding-affinity and specificity as the parental murine H23 mAb. Determination of the apparent binding affinities of the two antibodies from the half-maximal binding signal in ELISA and by whole-cell ELISA revealed that both antibodies bind

56

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

MUC1 with similar affinity (Fig. 3C, D). Furthermore, when tested for cellular MUC1 binding using flow cytometry, both antibodies stained similarly several MUC1 positive breast tumor cells that vary in MUC1 expression level. These results indicated that indeed conversion of the antibody from the monovalent format as was the case in our previous study, (Mazor et al., 2005) to a bivalent format in the form of chimeric H23 IgG1 restored the antibody binding avidity in more then a hundred-fold, to a level roughly equal to that of the parental murine H23 mAb. Next we evaluated the internalization capabilities of chH23 IgG1 by analyzing the ability of the purified antibody to internalize into MUC1 positive human breast carcinoma T47D cells using confocal microscopy. The results showed that no detectable internalization of chH23 could be observed. It was previously reported that some anti-Muc1 antibodies do internalized into Muc1-expressing cells while others do not (Pietersz and McKenzie, 1992; Henderikx et al., 2002; Krauss et al., 2004; Hamann et al., 2005a,b; Pericleous et al., 2005). In our case, when the chH23 IgG was evaluated for internalization into MUC1 positive tumor cells using confocal microscopy, no detectable internalization could be observed. This led us to hypothesize that perhaps chH23 does internalize into the MUC1 cells, however, not in a sufficient amount to be detected by means of fluorescent microscopy. This led us to further hypothesize that conjugation of chH23 IgG1 to a very potent toxin, where only a few internalized molecules are sufficient for killing of a target cell will shed light on the potential of this antibody to serve as a therapeutic tool for the treatment of human breast cancer. Attempts of arming chH23 with toxins or cytotoxic drug elements using standard methods of chemical conjugation through free NH2 groups were futile, as the antibody lost most its binding capability following such conjugation (unpublished results). This probably occurred because the CDR2 of H23 VH domain contains two lysine residues that presumably take part in antigen recognition and may be vulnerable to amine conjugation reagents (Benhar et al., 1994). We decided to take advantage of the IgG Fc-binding protein ZZ, which is a synthetic analogue of the immunoglobulin G (IgG)binding B domain of S. aureus protein-A (Nilsson et al., 1987). This recombinant protein with a total size of 12 kDa is well capable of binding the Fc domain of human IgG1 antibodies (Jansson et al., 1998) but much weaker to murine IgG1 (Nagaoka and Akaike, 2003). We describe here the construction of a recombinant fusion protein consisting of the potent Pseudomonas exotoxin A derivative (PE38) (Kreitman et al., 1992)

fused to the C-terminus of the ZZ domain and the linking of the ZZ-PE38 fusion protein to the chH23 antibody by the affinity connection; ZZ to the Fc domain. The resulting chH23-ZZ-PE38 immunocomplex with a total size of ∼ 200 kDa was separated of excess ZZ-PE38 by size-exclusion chromatography. Evaluation of the binding capabilities of chH23-ZZPE38 immunocomplex was performed by whole-cell ELISA, showing that the affinity tagging of the ZZPE38 fusion protein to chH23 Fc domain was stable, furthermore, the conjugation process did not undermine chH23 antigen recognition as the immunoconjugate retained its binding capabilities to MUC1 positive tumor cells with comparable affinity to the un-conjugated chH23 IgG1. Moreover, we demonstrated the specificity of chH23-ZZ-PE38 immunoconjugate to cellular MUC1, as binding of un-conjugated chH23 IgG1. The cytotoxic activity of chH23-ZZ-PE38 immunotoxin was evaluated by cell-killing experiments were the sensitivity of two MUC1 positive human breast carcinoma cell lines was tested. Both the highly MUC1-expressing T47D cells and the moderate MUC1-expressing MCF7 cells showed almost no sensitivity to the chH23-ZZPE38 immunocomplex. The results of this assay pointed out that chH23-ZZ-PE38 was not cytotoxic to the MUC1-expressing tumor cells; since sensitivity of reported immunotoxin-targeted cells represented IC50 values at the range of 0.1–10 ng/ml (Pai et al., 1991; Hassan et al., 2000). Pseudomonas exotoxin A (PE38) is a very potent toxin in which, once delivered to the cytosol, a single molecule is sufficient to kill a cell, however, in order to execute it cytotoxic activity it requires efficient internalization of the carrier antibody it is coupled to following specific binding to the target cell. This indicated that the affinity tagging of the ZZPE38 fusion protein to chFRP5 Fc domain did not undermine chFRP5 antigen recognition capability. Our results suggest that chH23-ZZ-PE38 was not cytotoxic to the MUC1-expressing tumor cells, probably since the immunocomplex does not internalize into the cells. From the results of the internalization experiments and our sensitive cell-killing analysis, it seems most likely that chH23 IgG1 is probably not an internalizing antibody. Nevertheless, it may be possible that the intracellular routing and processing of the MUC1receptor/toxin complex may affect the availability of the toxin counterpart to be released into the cytosolic compartment were it catalyzes the ADP-ribosylation and inactivation of elongation factor 2, which inhibits protein synthesis and leads to cell death. Still, chH23 has a high affinity and excellent specificity for human breast cancer cells. Thus to further explore its potential to serve

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

as a therapeutic tool for the treatment of human breast cancer, and since it does not activate complementmediated lysis or antibody-directed cytotoxicity as a naked antibody (unpublished data), it will probably be reasonable to couple this antibody to toxic effector molecules that do not require internalization. Such molecules could be radioisotopes as used in targeted radioimmunotherapy (Milenic et al., 2004). Indeed, clinical trials using radiolabeled anti-MUC1 antibodies for imaging and therapy were carried out or are currently in progress (DeNardo et al., 1997; Biassoni et al., 1998; Garkavij et al., 2005). Another strategy that may deal with both the penetration and the antigen heterogeneity problem is the antibody-directed enzyme therapy method (ADEPT) (Senter and Springer, 2001; Sharma et al., 2005). This is a two-step process for drug delivery in which monoclonal antibodies localize enzymes to tumor cell surface. The enzymes are selected for their abilities to convert subsequently administrated anticancer pro-drugs into active anti-tumor agents (Wu and Senter, 2005). To validate that the failure of the chH23-ZZ-PE38 immunocomplex to kill MUC1 tumor cells did not result from inactivity of the novel ZZ-PE38 fusion protein, we carried out experiments using the anti-ErbB2 antibody FRP5, whose capability to internalize and deliver a cytotoxic cargo into ErbB2 cells is well documented (Harwerth et al., 1992; Wels et al., 1992). We prepared a chimeric FRP5 derivative using our expression system described above and made the chFRP5-ZZ-PE38 immunocomplex. The chimeric FRP5 IgG1 and the derived immunocomplex bound avidly to target cells, but, in contrast to the chH23-ZZ-PE38 immunocomplex, the chFRP5-ZZ-PE38 immunocomplex also killed ErbB2-expressing breast cancer cells (Fig. 9). Among these cells were T47D and MCF7 that were not affected by the chH23-ZZ-PE38 immunocomplex, showing that the target cells themselves are not refractory to ZZPE38-based immunocomplex killing. Our results also indicated that chFRP5-ZZ-P38 was cytotoxic for all five cell lines tested with IC50 values that in most cases correlated with the levels of ErbB2 expression among the different tumor cells (Table 2). Taken together, the cell-killing activities achieved with chFRP5-ZZ-PE38 are a consequence of specific targeting and internalization of the immunocomplex to the tumor cells as none of the separable components alone significantly was toxic towards any of the tumor cells as well as the negative control hIgG-ZZ-PE38. Although we could not use antibody H23 to deliver a toxin to MUC1-expressing cells, we could demonstrate killing of such cells using a complex made by mixing

57

ZZ-PE38 with polyclonal serum obtained from mice that were immunized with recombinant MUC1 (Rubinstein et al., 2006). Clearly, using MUC1 as a target for toxin delivery is still a valid option. To conclude, ZZ-PE38 could prove to be a very effective tool for the evaluation of the therapeutic potential of antibodies and antibody targets. It could also provide for very effective targeted therapy, as we found by carrying out nude mouse xenograft studies with chFRP5-ZZ-PE38 (unpublished data). The IgG-ZZPE38 complex, although not covalent, is quite stable and the kinetics of competition between the resident IgG to external ones (such as the ones present in serum) are quite slow (unpublished results) and do not undermine studies of cell-killing in culture, as the ones shown here. Such competition will be of concern when in-vivo studies are done, and solutions to that problem can be provided by chemical crosslinking and additional approaches (unpublished data). Experiments are in progress to evaluate ZZ-PE38 in complex with other antibodies and antibody targets. Acknowledgments We thank Professor Winfried Wels (Georg Speyer Haus, Institute for Biomedical Research, Frankfurt am Main Germany) for providing the FRP5 carrying plasmid pHEN1-FRP5. We thank Dr. Ira Pastan (NCI, NIH, USA) for the anti-Pseudomonas exotoxin serum. We thank Dr. Orit Sagi-Assif (Tel-Aviv University) for assistance with FACS experiments. Supported in part by a grant from the Federico Foundation for Tel-Aviv University. References Ames, R.S., Tornetta, M.A., Deen, K., Jones, C.S., Swift, A.M., Ganguly, S., 1995. Conversion of murine Fabs isolated from a combinatorial phage display library to full length immunoglobulins. J. Immunol. Methods 184, 177. Azriel-Rosenfeld, R., Valensi, M., Benhar, I., 2004. A human synthetic combinatorial library of arrayable single-chain antibodies based on shuffling in vivo formed CDRs into general framework regions. J. Mol. Biol. 335, 177. Baruch, A., Hartmann, M., Yoeli, M., Adereth, Y., Greenstein, S., Stadler, Y., Skornik, Y., Zaretsky, J., Smorodinsky, N.I., Keydar, I., Wreschner, D.H., 1999. The breast cancer-associated MUC1 gene generates both a receptor and its cognate binding protein. Cancer Res. 59, 1552. Bebbington, C.R., Renner, G., Thomson, S., King, D., Abrams, D., Yarranton, G.T., 1992. High-level expression of a recombinant antibody from myeloma cells using a glutamine synthetase gene as an amplifiable selectable marker. Biotechnology (N Y) 10, 169. Benhar, I., Pastan, I., 1995. Characterization of B1(Fv)PE38 and B1 (dsFv)PE38: single-chain and disulfide stabilized Fv immunotoxins

58

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59

with increased activity that cause complete remissions of established human carcinoma xenografts in nude mice. Clin. Cancer Res. 1, 1023. Benhar, I., Brinkmann, U., Webber, K.O., Pastan, I., 1994. Mutations of two lysine residues in the CDR loops of a recombinant immunotoxin that reduce its sensitivity to chemical derivatization. Bioconjug. Chem. 5, 321. Benz, C.C., Scott, G.K., Santos, G.F., Smith, H.S., 1989. Expression of c-myc, c-Ha-ras1, and c-erbB-2 proto-oncogenes in normal and malignant human breast epithelial cells. J. Natl. Cancer Inst. 81, 1704. Biassoni, L., Granowska, M., Carroll, M.J., Mather, S.J., Howell, R., Ellison, D., MacNeill, F.A., Wells, C.A., Carpenter, R., Britton, K.E., 1998. 99mTc-labelled SM3 in the preoperative evaluation of axillary lymph nodes and primary breast cancer with change detection statistical processing as an aid to tumour detection. Br. J. Cancer 77, 131. Blake, R.A., 2001. Cellular screening assays using fluorescence microscopy. Curr. Opin. Pharmacol. 1, 533. Casalini, P., Caldera, M., Canevari, S., Menard, S., Mezzanzanica, D., Tosi, E., Gadina, M., Colnaghi, M.I., 1993. A critical comparison of three internalization assays applied to the evaluation of a given mAb as a toxin-carrier candidate. Cancer Immunol. Immunother. 37, 54. DeNardo, S.J., Kramer, E.L., O'Donnell, R.T., Richman, C.M., Salako, Q.A., Shen, S., Noz, M., Glenn, S.D., Ceriani, R.L., DeNardo, G.L., 1997. Radioimmunotherapy for breast cancer using indium-111/yttrium-90 BrE-3: results of a phase I clinical trial. J. Nucl. Med. 38, 1180. Durocher, Y., Perret, S., Kamen, A., 2002. High-level and highthroughput recombinant protein production by transient transfection of suspension-growing human 293-EBNA1 cells. Nucleic Acids Res. 30, E9. FitzGerald, D.J., Willingham, M.C., Pastan, I., 1988. Pseudomonas exotoxin-immunotoxins. Cancer Treat. Res. 37, 161. FitzGerald, D.J., Kreitman, R., Wilson, W., Squires, D., Pastan, I., 2004. Recombinant immunotoxins for treating cancer. Int. J. Med. Microbiol. 293, 577. Garkavij, M., Samarzija, M., Ewers, S.B., Jakopovic, M., Tezak, S., Tennvall, J., 2005. Concurrent radiotherapy and tumor targeting with 111In-HMFG1-F(ab')2 in patients with MUC1-positive nonsmall cell lung cancer. Anticancer Res. 25, 4663. Hamann, P.R., Hinman, L.M., Beyer, C.F., Greenberger, L.M., Lin, C., Lindh, D., Menendez, A.T., Wallace, R., Durr, F.E., Upeslacis, J., 2005a. An anti-MUC1 antibody-calicheamicin conjugate for treatment of solid tumors. Choice of linker and overcoming drug resistance. Bioconjug. Chem. 16, 346. Hamann, P.R., Hinman, L.M., Beyer, C.F., Lindh, D., Upeslacis, J., Shochat, D., Mountain, A., 2005b. A calicheamicin conjugate with a fully humanized anti-MUC1 antibody shows potent antitumor effects in breast and ovarian tumor xenografts. Bioconjug. Chem. 16, 354. Harwerth, I.M., Wels, W., Marte, B.M., Hynes, N.E., 1992. Monoclonal antibodies against the extracellular domain of the erbB-2 receptor function as partial ligand agonists. J. Biol. Chem. 267, 15160. Hassan, R., Viner, J.L., Wang, Q.C., Margulies, I., Kreitman, R.J., Pastan, I., 2000. Anti-tumor activity of K1-LysPE38QQR, an immunotoxin targeting mesothelin, a cell-surface antigen overexpressed in ovarian cancer and malignant mesothelioma. J. Immunother. 23, 473. Henderikx, P., Coolen-van Neer, N., Jacobs, A., van der Linden, E., Arends, J.W., Mullberg, J., Hoogenboom, H.R., 2002. A human immunoglobulin G1 antibody originating from an in vitro-selected

Fab phage antibody binds avidly to tumor-associated MUC1 and is efficiently internalized. Am. J. Pathol. 160, 1597. Hynes, N.E., Gerber, H.A., Saurer, S., Groner, B., 1989. Overexpression of the c-erbB-2 protein in human breast tumor cell lines. J. Cell Biochem. 39, 167. Hwang, W.Y., Foote, J., 2005. Immunogenicity of engineered antibodies. Methods 36, 3. Jansson, B., Uhlen, M., Nygren, P.A., 1998. All individual domains of staphylococcal protein A show Fab binding. FEMS Immunol. Med. Microbiol. 20, 69. Jostock, T., Vanhove, M., Brepoels, E., Van Gool, R., Daukandt, M., Wehnert, A., Van Hegelsom, R., Dransfield, D., Sexton, D., Devlin, M., Ley, A., Hoogenboom, H., Mullberg, J., 2004. Rapid generation of functional human IgG antibodies derived from Fabon-phage display libraries. J. Immunol. Methods 289, 65. Kim, J.H., Weaver, R.F., 1988. Construction of a recombinant expression plasmid encoding a staphylococcal protein A-ricin A fusion protein. Gene 68, 315. Krauss, J., Arndt, M.A., Zhu, Z., Newton, D.L., Vu, B.K., Choudhry, V., Darbha, R., Ji, X., Courtenay-Luck, N.S., Deonarain, M.P., Richards, J., Rybak, S.M., 2004. Impact of antibody framework residue VH-71 on the stability of a humanised anti-MUC1 scFv and derived immunoenzyme. Br. J. Cancer 90, 1863. Kreitman, R.J., Chaudhary, V.K., Siegall, C.B., FitzGerald, D.J., Pastan, I., 1992. Rational design of a chimeric toxin: an intramolecular location for the insertion of transforming growth factor alpha within Pseudomonas exotoxin as a targeting ligand. Bioconjug. Chem. 3, 58. Lang, P., Yeow, K., Nichols, A., Scheer, A., 2006. Cellular imaging in drug discovery. Nat. Rev. Drug Discov. 5, 343. Madshus, I.H., Stenmark, H., Sandvig, K., Olsnes, S., 1991. Entry of diphtheria toxin-protein A chimeras into cells. J. Biol. Chem. 266, 17446. Mazor, Y., Keydar, I., Benhar, I., 2005. Humanization and epitope mapping of the H23 anti-MUC1 monoclonal antibody reveals a dual epitope specificity. Mol. Immunol. 42, 55. Milenic, D.E., Brady, E.D., Brechbiel, M.W., 2004. Antibody-targeted radiation cancer therapy. Nat. Rev. Drug Discov. 3, 488. Nagaoka, M., Akaike, T., 2003. Single amino acid substitution in the mouse IgG1 Fc region induces drastic enhancement of the affinity to protein A. Protein Eng. 16, 243. Nilsson, B., Moks, T., Jansson, B., Abrahmsen, L., Elmblad, A., Holmgren, E., Henrichson, C., Jones, T.A., Uhlen, M., 1987. A synthetic IgGbinding domain based on staphylococcal protein A. Protein Eng. 1, 107. Nilsson, J., Larsson, M., Stahl, S., Nygren, P.A., Uhlen, M., 1996. Multiple affinity domains for the detection, purification and immobilization of recombinant proteins. J. Mol. Recognit. 9, 585. O'Hare, M., Brown, A.N., Hussain, K., Gebhardt, A., Watson, G., Roberts, L.M., Vitetta, E.S., Thorpe, P.E., Lord, J.M., 1990. Cytotoxicity of a recombinant ricin-A-chain fusion protein containing a proteolytically-cleavable spacer sequence. FEBS Lett. 273, 200. Pai, L.H., Batra, J.K., FitzGerald, D.J., Willingham, M.C., Pastan, I., 1991. Anti-tumor activities of immunotoxins made of monoclonal antibody B3 and various forms of Pseudomonas exotoxin. Proc. Natl. Acad. Sci. U. S. A. 88, 3358. Pastan, I., Kreitman, R.J., 2002. Immunotoxins in cancer therapy. Curr. Opin. Investig. Drugs 3, 1089. Pericleous, L.M., Richards, J., Epenetos, A.A., Courtenay-Luck, N., Deonarain, M.P., 2005. Characterisation and internalisation of recombinant humanised HMFG-1 antibodies against MUC1. Br. J. Cancer 93, 1257.

Y. Mazor et al. / Journal of Immunological Methods 321 (2007) 41–59 Pham, P.L., Perret, S., Doan, H.C., Cass, B., St-Laurent, G., Kamen, A., Durocher, Y., 2003. Large-scale transient transfection of serum-free suspension-growing HEK293 EBNA1 cells: peptone additives improve cell growth and transfection efficiency. Biotechnol. Bioeng. 84, 332. Pietersz, G.A., McKenzie, I.F., 1992. Antibody conjugates for the treatment of cancer. Immunol. Rev. 129, 57. Reiter, Y., Pastan, I., 1998. Recombinant Fv immunotoxins and Fv fragments as novel agents for cancer therapy and diagnosis. Trends Biotechnol. 16, 513. Rubinstein, D.B., Carmeli, M., Ziv, R., Benhar, I., Leitner, O., Baron, S., Katz, B.-Z., Wreschner, D.H., 2006. MUC1/X protein immunization enhances cDNA immunization in generating antiMUC1 alpha/beta junction antibodies that target malignant cells. Cancer Res. 66, 11247. Sanna, P.P., 2002. Expression of antibody Fab fragments and whole immunoglobulin in mammalian cells. Methods Mol. Biol. 178, 389. Sauer, P.W., Burky, J.E., Wesson, M.C., Sternard, H.D., Qu, L., 2000. A high-yielding, generic fed-batch cell culture process for production of recombinant antibodies. Biotechnol. Bioeng. 67, 585. Schmidt, M., Hynes, N.E., Groner, B., Wels, W., 1996. A bivalent single-chain antibody-toxin specific for ErbB-2 and the EGF receptor. Int. J. Cancer 65, 538. Schrama, D., Reisfeld, R.A., Becker, J.C., 2006. Antibody targeted drugs as cancer therapeutics. Nat. Rev. Drug Discov. 5, 147. Senter, P.D., Springer, C.J., 2001. Selective activation of anticancer prodrugs by monoclonal antibody-enzyme conjugates. Adv. Drug Deliv. Rev. 53, 247. Sharma, S.K., Bagshawe, K.D., Begent, R.H., 2005. Advances in antibody-directed enzyme prodrug therapy. Curr. Opin. Investig. Drugs. 6, 611. Simmons, L.C., Reilly, D., Klimowski, L., Raju, T.S., Meng, G., Sims, P., Hong, K., Shields, R.L., Damico, L.A., Rancatore, P., Yansura,

59

D.G., 2002. Expression of full-length immunoglobulins in Escherichia coli: rapid and efficient production of aglycosylated antibodies. J. Immunol. Methods 263, 133. Tonevitskii, A.G., Rakhmanova, V.A., Toptygin, A., Zdanovskii, A.G., Ershova, G.V., Zdanovskaia, M.V., Rutkevich, N.M., Iankovskii, N.K., 1991. Cytotoxic properties of a recombinant hybrid of the A protein of Staphylococcus aureus with a fragment of exotoxin A of Pseudomonas aeruginosa. Mol. biol. ( Mosk. ) 25, 1188. Trill, J.J., Shatzman, A.R., Ganguly, S., 1995. Production of monoclonal antibodies in COS and CHO cells. Curr. Opin. Biotechnol. 6, 553. Tsarfati, I., Hareuveni, M., Choo, C.S., Horev, M., Hizi, A., Wreshner, D.H., Keydar, I., 1989. H23 monoclonal antibodies recognize a breast tumor associated antigen: clinical and molecular studies. Monoclonal Antibodies and Breast Cancer. Martinus Nijhoff Publishing, San Francisco. Tsurushita, N., Hinton, P.R., Kumar, S., 2005. Design of humanized antibodies: from anti-Tac to Zenapax. Methods 36, 69. Wels, W., Harwerth, I.M., Mueller, M., Groner, B., Hynes, N.E., 1992. Selective inhibition of tumor cell growth by a recombinant singlechain antibody-toxin specific for the erbB-2 receptor. Cancer Res. 52, 6310. Wu, A.M., Senter, P.D., 2005. Arming antibodies: prospects and challenges for immunoconjugates. Nat. Biotechnol. 23, 1137. Xie, L., Wang, D.I., 1997. Integrated approaches to the design of media and feeding strategies for fed-batch cultures of animal cells. Trends Biotechnol. 15, 109. Yacoby, I., Shamis, M., Bar, H., Shabat, D., Benhar, I., 2006. Targeting antibacterial agents by using drug-carrying filamentous bacteriophages. Antimicrob. Agents Chemother. 50, 2087. Yoo, E.M., Chintalacharuvu, K.R., Penichet, M.L., Morrison, S.L., 2002. Myeloma expression systems. J. Immunol. Methods 261, 1.