Antibody Internalization after Cell Surface Antigen Binding is Critical

Sep 28, 2009 - Immunotoxin potency is dependent on cell surface binding specificity ... approaches for immunotoxin development are dependent on existi...
0 downloads 0 Views 2MB Size
Bioconjugate Chem. 2009, 20, 1975–1982

1975

Antibody Internalization after Cell Surface Antigen Binding is Critical for Immunotoxin Development Shu-Ru Kuo, Randall W. Alfano, Arthur E. Frankel, and Jen-Sing Liu* Cancer Research Institute of Scott & White Hospital; Department of Internal Medicine, Health Science Center, College of Medicine, Texas A&M University, 5701 South Airport Road, Temple, Texas 76502. Received July 27, 2009; Revised Manuscript Received September 3, 2009

Immunotoxin potency is dependent on cell surface binding specificity as well as internalization efficiency. Current approaches for immunotoxin development are dependent on existing antibodies that were selected for high affinity and/or high production yield. However, these antibodies may demonstrate low internalization efficiency upon cell surface binding and thus are not necessarily the best candidates for immunotoxin design. Here, we have developed an assay with a novel protein, DTG3, to compare and evaluate the internalization efficiency of monoclonal antibodies in order to circumvent the possibility of low internalization. DTG3 is a fusion protein containing the N-terminus of diphtheria toxin (DT) and three copies of streptococci Protein G immunoglobulin binding domains. We show that antibody-DTG3 complexes formed in the test tube are able to bind their antigen on the target cell surface, resulting in cell internalization, DT-mediated protein synthesis inhibition, and host cell apoptosis. We tested this system with two well-studied antibodies, antihuman CD3ε, and anti-PSMA antibodies and were able to show efficiency of this assay. We further examined commercially available anti-CD123 antibodies for potential leukemia-targeting immunotoxin development. Finally, we applied this system in the early-stage screening of newly generated anti-CD123 hybridomas. Our data showed that this internalization assay system is sensitive, time efficient, and reproducible, and has provided a tool to compare monoclonal antibodies for the clinical development of effective immunotoxins for the treatment of a variety of neoplasms.

INTRODUCTION Attempts to explicitly target proliferating tumor cells, and thus enhance therapeutic efficacy, have led to the development of targeted immunotoxins (1). These agents direct the powerful lethal action of recombinant bacterial or plant toxins to tumor cells through the fusion to growth factors, monoclonal antibodies, or antibody fragments that recognize tumor cell-specific epitopes. Toxins typically utilized for this application are referred to AB-toxins due to their structural organization. The A moiety consists of an enzymatic domain that modifies a specific intracellular target, causing the disruption of cell functions that are critical for cell survival. The B moiety, which is often modified for tumor cell targeting, comprises one or multiple subunits that bind cell surface receptors and permit the translocation of the A moiety into the host cytosol. Rapid progress in the understanding of toxin structure and function has fueled the development of numerous novel immunotoxins for use in hematological cancer therapy (2). The selection of cell-binding peptides that have high specificity to the intended target cell as well as high internalization efficiency upon cell surface binding is critical in immunotoxin design. Traditional methods used to study internalization efficiency of cell surface targeting peptides were dependent on the use of isotopes such as 125I or microscopy techniques (3-6). However, these methods proved to have inadequate sensitivity. More sensitive techniques included chemically conjugating the cell-targeting antibody to a toxin to measure cell killing directly (7). Since catalytic toxins are able to kill cells with as few as one molecule per cell, this assay can have much higher * To whom correspondence should be addressed. Dr. Jen-Sing Liu, Cancer Research Institute of Scott & White, 5701 South Airport Road, Temple, Texas 76502, USA. Tel: (254) 724-1839; Fax: (254) 7242324; Email: [email protected].

sensitivity than isotope or fluorescent-based techniques. However, chemical conjugation is time-consuming, costly, and has poor reproducibility. Recently, Mazor et al. (8) reported the development of a novel IgG binding toxin fusion protein (ZZPE38) for antibody internalization assays. The ZZ-PE38 fusion protein contains two copies of IgG-binding motif from Streptococcal protein A and a truncated Pseudomonas exotoxin A (PE38). The strong binding of protein A to the IgG Fc domain generated a bridge between the IgG and the catalytic toxin and, therefore, conveniently eliminated the need for chemical conjugation. Purified ZZ-PE38 and antibody complexes were shown to be homogeneous with antigen-binding domain exposed. This method has proved to be an invaluable tool to be further developed for antibody internalization studies. Our laboratory has been studying the therapeutic potential of several diphtheria toxin based immunotoxins for treatment in leukemia (9). Efforts have been focused on targeting the interleukin-3 (IL3)1 receptor via a fusion protein composed of the catalytic and translocation domains of diphtheria toxin (DT388) linked to the IL3 cytokine (10-13). This fusion protein, known as DT388IL3, is currently under evaluation for efficacy and safety in phase I clinical trials (9). Although still ongoing, alternative approaches for IL3R targeting are already underway. The R subunit of IL3R (CD123) has been shown to be up-regulated in acute myeloid leukemias, acute lymphoblastic leukemias, plasmacytoid dendritic cell leukemias, and some chronic myeloid leukemias (3, 14-17). In contrast, the IL3R common β subunit (CDw131) is not always coexpressed in those leukemias (18). Thus, therapeutic agents targeting CD123 alone have potential for broader applications. 1 Abbreviations: IL3, interleukin-3; IL3R, interleukin-3 receptor; IgG, immunoglobulin G; DT, diphtheria toxin; PE, Pseudomonas exotoxin; AML, acute myeloid leukemia; PSMA, prostate specific membrane antigen; DMP, dimethyl pimelimidate; PBS, phosphate buffer saline.

10.1021/bc900333j CCC: $40.75  2009 American Chemical Society Published on Web 09/28/2009

1976 Bioconjugate Chem., Vol. 20, No. 10, 2009

These findings prompted the development of immunotoxins targeting CD123. Du et al. (19) produced constructs consisting of Pseudomonas exotoxin PE38 and single-chain Fv of antiCD123 antibodies. In Vitro potency assays showed that an optimized construct, 26292(Fv)-PE38KDEL, had an IC50 in leukemia cells at around 500 pM. Although encouraging, this construct exhibited lower cytotoxicity than that of DT388IL3 (IC50 at ∼10 pM) (12, 20). Since 26292(Fv)-PE38KDEL showed strong cell-surface binding to the targeted cells, the observed 50-fold lower potency might have been a result of low internalization efficiency. In order to address these preliminary findings, we have developed a new antibody internalization assay with a novel recombinant protein, DTG3. This protein consists of the truncated diphtheria toxin (DT388) fused to 3 copies of IgG biding motif from group G streptococci Protein G (21). Here, we describe the efficiency of this method using the well-studied antihuman CD3ε and PSMA antibodies in order to demonstrate biological activity and thus provide proof of principle. Our results indicate that the DTG3 coupling system described here provides a rapid method with high sensitivity to compare and evaluate antibody internalization efficiencies. We further prove that this method can be used in the early stage of hybridoma screening to identify new anti-CD123 antibodies with high internalization efficiency.

EXPERIMENTAL PROCEDURES Plasmids. (1) pDTG1-His, pDTG2-His, and pDTG3-His: The plasmids expressing DT-proteinG fusion proteins were made of 2 parts: the sequence of the N-terminal 388 amino acids of diphtheria toxin (DT) was derived from pRKDTIL3 (22); and the DNA fragment containing the codon-maximized sequence for IgG binding proteins and linkers was synthesized by GenScript (Piscataway, NJ). To construct IgG binding protein, the sequence of the second IgG binding domain of wild-type group G streptococci Protein G (GBII) (21) was used. Seven and 8 amino acids immediately adjacent to the N-terminus and C-terminus, respectively, of the 55-amino-acid GBII core domain were also added to provide spacers when multiple copies were cloned. This amino acid sequence was reverse translated into the DNA sequence following preferred codon usage in E. coli. Six-nucleotide linkers recognized by restriction enzyme BamH 1 and BglII were added to the 5′ and 3′ ends, respectively. To the 3′ end, DNA sequence of 6 histidines, a termination codon (TGA) and a Hind III recognition site were then added. The entire sequence was synthesized by GenScript and delivered within pUC57 plasmid (pUC57-G1-His). The BamH 1-Hind III fragment of pUC57-G1-His was inserted to pET-EE plasmid (23) to generate pET-EE-G1-His. The BamH 1-BglII fragment from pUC57-G1-His was then inserted to BamH1 site of pETEE-G1-His to produce pET-EE-G2-His. The same strategy was used to generate pET-EE-G3-His. The Nde I-Hind III fragments from pET-EE-G1-His, pET-EE-G2-His, and pET-EE-G3-His were inserted into pRKDTIL3 to create pDTG1-His, pDTG2His, and pDTG3-His expression plasmids. (2) pcDNA-CD123: The cDNA of human CD123 was reverse-transcribed and amplified from leukemia cell line TF1 using Qiagen RNeasy mini and onestep RT-PCR kits. The PCR primers used are 5′CCGGAGCTAGCGTTCCCGATGGTCC-3′ with one underlined Nhe I recognition site and 5′-TGGCCCGGCCGGGGAGATTGAGGCA-3′. The 1.2 kb PCR product was first cloned into the pDrive vector using Qiagen PCR cloning kit and followed by subcloning into the KpnI and Nhe I sites of a eukaryotic expression vector pcDNA3.1/Zeo (Invitrogen, Carlsbad, CA) to generate pcDNA-CD123. Cell Lines and Antibodies. Jurkat cells, kindly provided by Dr. David M. Neville, and TF1/H-ras (24) cells were maintained

Kuo et al.

in RPMI medium containing 10% FBS. PSMA-stably transfected NIH3T3 cells, a kind gift from Dr. Michel Sadelain, were maintained in DMEM medium containing 10% FBS. CHO-K1 cells (ATCC, Manassas, VA) were maintained in ATCCformulated F-12K medium containing 10% FBS. pcDNACD123-stably transfected CHO-K1 cells expressing high copy number of CD123 were cultured in F-12K medium supplemented with 10% FBS and 0.4 mg/mL zeocin (Invitrogen, Carlsbad, CA). Antihuman CD3ε antibody UCHT1 and antihuman PSMA antibody J591 were kindly provided by Drs. Jung Hee Woo and Neil H. Bander. Anti-CD123 antibodies 7G3 and 6H6 were purchased from BD Biosciences (San Jose, CA) and Abcam (Cambridge, MA), respectively. Anti-CDw131 antibody 1C1 was purchased from eBioscience (San Diego, CA). Expression and Purification of DTG1, DTG2, and DTG3. Bacteria BL21(DE3) transformed with pDTG1-His, pDTG2-His, or pDTG3-His were cultured in SuperBroth containing 1.6 mM MgCl2, 0.5% glucose, and 100 µg/mL ampicillin. A 4 mL overnight culture was added to 200 mL fresh medium at 37 °C until OD600 reached 0.6. The culture was cooled down to 18 °C (∼1 h) before addition of IPTG to a final concentration of 0.2 mM. The induction was continued at 18 °C for 18 h. The cells were harvested and treated with 100 µg/mL lysozyme on ice for 30 min. The cell pellet was then resuspended in 20 mL of ice-cold hypotonic buffer (20 mM HEPES-K+ pH 7.5, 5 mM KCl, 1 mM MgCl2) on ice for 5 min. Cells were lysed by sonication once at 35 W for 30 s, followed by the addition of 5 mL 5 M NaCl, and three more sonications at 35 W for 30 s on ice. After centrifugation at 15 000 × g for 10 min, clear supernatant was mixed with 1 mL of NiCl2-charged Chelator Sepharose beads (GE Healthcare, Piscataway, NJ) at 4 °C for 2 h. Beads were spun down and washed with 20 mM HEPES-K+ pH 7.5 with 1 M NaCl followed by 5 mL each of Buffer A (20 mM HEPES-K+ pH 7.5 with 100 mM NaCl), Buffer A with 2 mM imidazole, and Buffer A with 20 mM imidazole. The bound proteins were eluted with Buffer A with 200 mM imidazole. The collected protein peak fraction (about 1 mL) was loaded immediately into a 10/30 Superose 200 column (GE Healthcare, Piscataway, NJ) pre-equilibrated with Buffer A using a Purifier system (GE Healthcare, Piscataway, NJ). The fractions containing DTproteinG fusion proteins (about 1.5 mL) was pooled, aliquoted, and stored at -20 °C. Protein Gel Analysis. In a 10 µL reaction, 1 µg of DTG1, DTG2, or DTG3 was mixed with 0, 2, or 4 µg of antibody in phosphate buffer saline (PBS) pH 7.4 at room temperature. Control reactions had only 4 µg of antibody. For native gel analysis, the mixture was incubated at room temperature for 60 min. Two microliter aliquots of loading buffer (20 mM TrisHCl pH 8.0, 50% glycerol) were added to each reaction before loading to a 5% native polyacrylamide gel. Electrophoresis was carried out at 60 V for 4 h on ice. The gel was then fixed with 35% methanol and 15% acetic acid for 10 min followed by Coomassie blue staining (BioRad, Hercules, CA) for 30 min. After destaining, images were taken by a gel documentation system and labeled by Photoshop. For denatured gel analysis, the same reactions as described above were incubated at room temperature for 45 min before the addition of 1 µL 200 mM freshly prepared dimethyl pimelimidate (DMP; Sigma-Aldrich, St. Louis, MO) in PBS. The reactions were incubated for another 15 min before the addition of 1 µL of 1 M Tris-HCl pH 8.0 to neutralize unused DMP. Four microliters of 4× SDS sample buffer (Invitrogen, Carlsbad, CA) were added into each reaction and analyzed by an 8% SDS polyacrylamide gel running at 120 V for 90 min.

Antibody Internalization Assays

The gel was stained and destained as described above. Images were taken by a gel documentation system and labeled by Photoshop. Chemical Conjugation. Antibodies were conjugated to CRM9 (a mutated DT) following the protocol of ZangemeisterWittke (7). Ten milligrams of nicked CRM9 was prepared at 5 mg/mL in 50 mM sodium borate pH 8.0 with 1 mM EDTA. A fresh solution of SMCC (Pierce, Rockford, IL) was prepared at a concentration of 10 mM in dry DMF (Sigma-Aldrich, St. Louis, MO). A 20-fold molar excess of SMCC (Pierce, Rockford, IL) over CRM9 was added and mixed for 1 h at room temperature. This would add maleimide groups to CRM9 proteins. The filtered mixture was then added to a PD-10 desalting column (GE Healthcare, Piscataway, NJ) equilibrated with nitrogen flushed 10 mM sodium phosphate pH 7.0. The first peak was collected according to Bradford reading (BioRad, Hercules, CA), which corresponds to the derivitized toxin. The derivitized CRM9 was kept at 4 °C until use. Two milligrams of antibodies was prepared in 50 mM sodium borate pH 8.0 and 1 mM EDTA. They were derivitized by adding a 20-fold molar excess of fresh 10 mM 2-IT (iminothiolane) solution (Pierce, Rockford, IL) in borate buffer, pH 8.3. This would add terminal sulfhydryl groups to antibodies. The mixture was allowed to react at room temperature for 1 h before it was applied to a PD-10 column (GE Healthcare, Piscataway, NJ) equilibrated with nitrogen-flushed 10 mM sodium phosphate pH 7.0. The first protein peak was collected and added to the derivitized CRM9 in a 15 mL conical tube. The mixture was allowed to react for 20 h at room temperature. Conjugates were purified through a Superdex 200 column (GE Healthcare, Piscataway, NJ) equilibrated by 10 mM sodium phosphate pH 8.0, 90 mM sodium sulfate, and 1 mM EDTA. The peak corresponding to the monoconjugated antibody was subsequently collected and concentrated using the centrifugal filter (Millipore, Billerica, MA). Concentration was determined via Bradford assay (BioRad, Hercules, CA). Complex Formation. In a 10 µL reaction, 4 µg of monoclonal antibody was mixed with 2.5 µg of DTG3 at room temperature for 45 min. One microliter of 200 mM freshly prepared DMP in PBS was then added into each reaction and incubated for another 15 min before the addition of 1 µL 1 M Tris-HCl pH 8.0 to neutralize unused DMP. For the control, either antibody or DTG3 was omitted from the reactions. Potency Assays. Potency of antibodies complexed with DTproteinG fusion proteins or antibody-CRM9 conjugates was evaluated by their ability to suppress 50% of H3-thymidine incorporation in treated cells (IC50). The reactions were first diluted with fresh culture medium to the highest testing concentration (from 3 to 100 nM) based on antibody contents into the first column of 96-well round-bottom plates. Each sample was either triplicated or quadruplicated in the same plate. The first column was then diluted 1:3 to the second column and 3-fold dilution continued all the way to the 12th column. 50 µL aliquots of the diluted complexes were added to 96-well cell plates containing 1 × 104 cells in 100 µL per well. The cells were incubated at 37 °C in CO2 incubator for 48 h before the addition of 1 µCi of H3-thymidine to each well. After additional 18 h incubation at 37 °C in CO2 incubator, the cells were harvested and counted in a β-counter. The potency curves were graphed using Prism fitted with sigmoid dose response curve with variable slopes. Hybridoma Cell Line Production. All experiments involving animals were preapproved by IACUC of Scott & White Hospital (2008-049-R). Briefly, groups of 3 Balb/c mice were intraperitoneally injected with 5 million CD123-transfected CHO-K1 cells at day 1, day 15, and day 29. Blood samples were collected at day 36 to evaluate antibody titer induced in

Bioconjugate Chem., Vol. 20, No. 10, 2009 1977

each animal by C-ELISA (see below) with parental or CD123transfected CHO-K1 cells. The mouse with higher antibody titer was given a final boost at day 57 with intravenous injection of 2 million CD123-transfected CHO-K1 cells. Splenocytes were harvested at day 60 and fused with P3 × 63Ag8.653 myeloma cells (ATCC, Manassas, VA) following the published protocol (25, 26). Hybridoma cells were mixed in HAT selection medium containing methylcellulose and Hybridoma Fusion and Cloning Supplement (HFCS; Roche, Basel, Switzerland) and plated in ten 100-mm Petri dish (26). Ten days later, single colonies were picked and expanded in 96-well flat bottom plates with 200 µL of hybridoma culture medium. Culture supernatant from each well was diluted 1:10 and tested in C-ELISA with parental or CD123-transfected CHO-K1 cells to select clones that produced anti-CD123 antibodies. The initial culture supernatant from positive clones was used for iso-type identification with an Iso-Gold kit (BioAssay Works, Ijamsville, MD) and for IgG quantification with a mouse-IgG ELISA kit (Roche, Basel, Switzerland). The identified anti-CD123 antibodyproducing hybridomas cell lines were expanded for further antibody purification. Cell-Based ELISA (C-ELISA). One day prior to the experiment, 5× 104 parental or CD123-transfeced CHO-K1 cells were seeded in each well of a flat-bottom 96-well plate. The next day, cells were washed with PBS and fixed with 3% formaldehyde in PBS at room temperature for 10 min. Fixed cells were washed thrice with PBS with 0.5% Tween 20 and incubated in Blocking Solution (PBS with 10% goat serum) for 30 min. Primary antibodies were then added at room temperature for 60 min followed by HRP-conjugated secondary antibodies (Jackson Lab, West Grove, PA) at room temperature for 45 min. The HRP activity was detected by HRP substrate (R&D, Minneapolis, MN) for 10 min and stopped by 0.5 M H2SO4. The plates were read by a plate reader set at 450 nm. Antibody Purification. Hybridoma culture supernatant (200 mL) was mixed with 400 mL of 20 mM Tris-HCl pH8.0 at room temperature and applied to a 5 mL Q-Sepharose column (GE Healthcare, Piscataway, NJ). Total bound proteins were eluted with 20 mM Tris-HCl pH 8.0 with 1 M NaCl, adjusted to a final NaCl concentration of 3 M, and applied to a 2 mL Protein G Sepharose column (Pierce, Rockford, IL). Antibodies were eluted with 100 mM glycine pH 3.0 and neutralized immediately with 1/10 volume of 1.5 M Tris-HCl pH 8.8. Desalting column PD-10 (GE Healthcare, Piscataway, NJ) was used to exchange the buffer of purified antibodies to PBS with 20% glycerol. Protein concentrations were measured by Bradford assays (BioRad, Hercules, CA) with BSA (Pierce, Rockford, IL) as standards and verified by reduced SDS-PAGE.

RESULTS Construction, Expression, and Production of DTG1, DTG2, and DTG3. DT-proteinG fusion proteins carrying one to three copies of the protein G IgG binding domains were expressed and purified from bacteria transformed with overexpression plasmids. These DT-ProteinG fusion proteins carry the N-terminus of diphtheria toxin (DT388) that contains the catalytic and translocation domains but not the cell surface-binding domain (Figure 1A). The purified proteins with expected sizes were g95% pure (Figure 1B). Complex Formation and Stabilization. The ability of the DT-ProteinG fusion proteins to form complexes with monoclonal antibodies was examined by native protein gel electrophoresis. Increasing amounts of anti-PSMA antibody J591 were titrated to fixed amounts of DT-proteinG fusion proteins. While both DTG2 and DTG3 formed stable complexes with J591 and appeared as a more slowly migrating band (compare lanes 6 and 9 to lane 10), DTG3 had a higher complex-forming

1978 Bioconjugate Chem., Vol. 20, No. 10, 2009

Kuo et al.

Figure 1. Construction and production of DTG1, DTG2, and DTG3. (A) N-terminal 388 amino acids of diphtheria toxin (DT388) was fused to a 9-amino-acid EE spacer (32) and 1-3 copies of the second copy of protein G IgG binding domain (GBII). The C-terminal histidine tag (6-His) allows easy purification. (B) Purified DTG1, DTG2, and DTG3 migrated in a 10% SDS-PAGE between markers of 50 and 75 kDa.

Figure 3. Potency of (A) UCHT1-CRM9 conjugates and (B) UCHT1DTG3 complexes in Jurkat cells.

Figure 2. J591 and DT-proteinG fusion protein complex formation. (A) Native gel analysis. One microgram of DTG1, DTG2, or DTG3 was mixed with 0, 2.5, or 5 µg of J591 in PBS pH 7.4 at room temperature for 1 h. Protein mixtures were then loaded to a 5% native polyacrylamide gel and run on ice at 60 V for 4 h. (B) SDS-PAGE analysis. The same reactions as in (A) were added with 20 mM DMP to stabilize the complexes and loaded to a nonreduced 8% SDS-PAGE. Migration of each protein monomer, complexes, and the sizes of protein markers are labeled on the sides of each diagram. The faster migration of J591 at ∼130 kDa (lane 10) was caused by DMP cross-linking.

efficiency (Figure 2A). The complete shift of 1 µg of DTG3 appeared at lane 8 where the antibody to DTG3 molar ratio was at about 1:1. Unfortunately, the protein complexes migrated only slightly more slowly than J591 alone, and J591-DTG1 complexes, if present, were not distinguishable from J591 (lanes 1-3). In order to separate J591 from the complexes, a bifunctional cross-linker dimethyl pimelimidate (DMP) was added to stabilize these complexes and then analyzed in SDS-PAGE (Figure 2B). While DMP cross-linked only ∼70% of the complexes (compare Figure 2A, lanes 8 and 9 to Figure 2B, lanes 9 and 10), this reagent does help to maintain the complex stability between J591 and DTG1 (Figure 2B, lanes 2 and 3). On the basis of the migration of the complexes to calculate their molecular weights,

J591 and DT-proteinG fusion proteins formed complexes at 1:1 ratio. We concluded that a single copy of GBII has the ability to form complex with antibodies while multiple copies of GBII can help to increase stability of the complexes. On the basis of these results, we chose DTG3 to further evaluate the system. Potency of UCHT1 and DTG3 Complexes. We subsequently determined the potency of DTG3/monoclonal antibody complexes. In order to do this, we chose the well-characterized immunotoxin A-dmDT390-bisFv(UCHT1) (27). This construct consists of diphtheria toxin fused to bivalent antihuman CD3ε antibody. The parental monoclonal antihuman CD3ε antibody UCHT1 was chemically conjugated to a cell surface bindingdeficient diphtheria toxin mutant CRM9 using a set of crosslinkers and subsequently purified (Materials and Methods). Potency of UCHT1-CRM9 was tested in Jurkat cells. The IC50 of two separated batches of UCHT1-CRM9 conjugates were at 4 pM and 13 pM (Figure 3A and data not shown). The DTG3 was then used to replace CRM9 in the same tests. UCHT1 was mixed with DTG3 in PBS in 10 µL reactions at room temperature. The complexes were stabilized by DMP. Without further purification, potency of these complexes was tested directly on Jurkat cells. While UCHT1 or DTG3 alone had no toxicity to treated cells, UCHT1 coupled with DTG3 had IC50 at 0.32 pM (Figure 3B). Potency of J591 and DT-Protein G Fusion Protein Complexes. The potential of using DTG3 to study antibody internalization was further evaluated in another ongoing project

Antibody Internalization Assays

Bioconjugate Chem., Vol. 20, No. 10, 2009 1979 Table 1. Antibodies Qualification and Quantification culture supernatant

purified antibody

clones

subtype

µg/mL

nM

IC50 (pM)

Kd

IC50 (pM)

12F1 12E8 12H12 12E7 16G7 7G3a 6H6a

IgG2bκ IgG2aκ IgG2bκ IgG2aκ IgG2aκ IgG2aκ IgG1κ

4.7 5.3 4.8 0.8 5.2

29 33 30 5 33

34 76 30 217 77

50 ( 1 65 ( 20 57 ( 17 63 ( 15 75 ( 29 131 ( 29 175 ( 43

42 ( 7 175 ( 11 42 ( 5 117 ( 2 120 ( 7 423 ( 66 620 ( 500

a 7G3 and 6H6 are commercial antibodies and the iso-type information was provided by venders.

Figure 4. Potency of J591-DTG3 complexes in PSMA-transfected NIH3T3 cells.

Figure 6. Cytotoxicity assay. Hybridoma culture supernatant coupled with DTG3 was tested in CD123-transfected CHO-K1 cells.

Figure 5. Potency of anti-IL3R antibodies-DTG3 complexes in CD123transfected CHO-K1 cells.

in our group. Anti-PSMA antibody J591 was shown to have a high internalization rate upon binding to cell-surface PSMA proteins (28). When J591 was mixed with DTG3 at room temperature and stabilized by DMP, the potency of the complexes was tested in PSMA-transfected NIH3T3 cells. As shown in Figure 4, while J591 or DTG3 alone and UCHT1DTG3 complexes had no toxicity to treated cells, J591-DTG3 complexes had an IC50 at 36 pM. Potency of rIL3R Antibodies and DT Fusion Protein Complexes. Our optimal goal is to evaluate the possibility of generating an anti-IL3R immunotoxin for leukemia treatment. Two most commonly used anti-IL3RR (CD123) antibodies 7G3 and 6H6 and anti-IL3Rβ (CDw131) antibody 1C1 were obtained commercially and mixed with DTG3 at room temperature. The complexes were stabilized by DMP. The potency of these complexes was then tested in CD123-transfected CHO-K1 cells. The IC50 for two different lots of 7G3 coupled with DTG3 was at 423 ( 66 pM (Figures 5 and 8, Table 1, and data not shown). Conversely, in repeated experiments, two different lots of 6H6 showed inconsistent results from 180 pM to 1.2 nM (Figures 5 and 8, Table 1, and data not shown) for unclear reasons. Regardless of their moderate potency as efficient immunotoxin candidates, they clearly showed that different clones of antibodies can have different internalization efficiency. Since there is no cell surface CDw131 (IL3Rβ) on CHO-K1 cells, 1C1 coupled with DTG3 showed only low toxicity at a very high immunotoxin concentration (Figure 5). Production of New Anti-CD123 Antibodies. To produce anti-CD123 antibodies that have higher internalization efficien-

cies, Balb/c mice were immunized with CD123-transfected CHO-K1 cells. Hybridoma cells were generated by fusing splenocytes from the immunized mice and myeloma cells. To identify the antibody producing hybridoma, the culture supernatants from individual clones were screened for their ability to bind CD123-transfected CHO-K1 cells but not parental CHOK1 cells in C-ELISA. These CHO-K1 cells were seeded in 96well plates and fixed with formaldehyde before the addition of antibodies to prevent potential internalization. Among ∼600 selected hybridoma clones from 1 immunized mouse, 5 clones (12F1, 12E8, 12H12, 12E7, 16G7) showed positive response. The specific binding was further confirmed by flow cytometry analysis. The cytometry graph showed that all five clones bind CD123-transfected CHO-K1 and leukemia cells TF1/H-ras but not parental CHO-K1 or lymphocyte Jurkat cells (data not shown). One hundred microliters of initial culture supernatant from these five positive clones were saved from the 96-well plates to be used in several assays. The first one is an immunoglobulin iso-typing assay. 12F1 and 12H12 were identified to be IgG2bκ; and 12E7, 12E8 and 16G7 were IgG2aκ (data not shown; Table 1). Mouse IgG ELISA quantification assays were then used to measure antibody concentrations in these crude culture supernatants. Four of the five clones had very similar antibody concentrations at 4-5 µg/mL or ∼30 nM, but slower-growing 12E7 had a much lower concentration (0.8 µg/mL; ∼5 nM) than the others (Table 1). Twenty microliters of each culture supernatant (80-100 ng, or 16 ng of 12E7) were mixed with an excess amount of DTG3 (200 ng) at room temperature for 1 h and stabilized with DMP. DTG3 alone at this dose range does not have any nonspecific cytotoxic effect (data not shown, but see Figure 8). Serial dilutions of these antibody-DTG3 mixtures were tested in potency assays with CD123-transfected CHO-K1 cells. As shown in Figure 6, different internalization rates were observed among these five clones. On the basis of the titration curve and

1980 Bioconjugate Chem., Vol. 20, No. 10, 2009

Figure 7. Protein quantification and qualification. (A) 2 µg of purified antibodies and commercial available 7G3 and 6H6 were analyzed by a reduced 12.5% SDS-PAGE. (B) Cell binding affinity of purified antibodies was tested in C-ELISA with CD123-transfected CHO-K1 cells.

individual concentration, DTG3-coupled 12F1 and 12H12 showed higher potency in CD123-transfected CHO-K1 cells with IC50 calculated to be at around 30-35 pM (Table 1). To further confirm our findings, these hybridoma cell lines were expanded for antibody production. Monoclonal antibodies were purified via protein-G chromatography (Figure 7A). Both 12F1 and 12H12 had higher levels of modifications, possibly glycosylation, on the heavy chains. C-ELISA was used again to examine their binding affinity (Figure 7B; Table 1). All five new anti-CD123 antibodies showed higher CD123 binding affinity than the commercially available 7G3 or 6H6. The same amounts of purified antibodies were then mixed with DTG3, stabilized by DMP and tested for their cytotoxicity with parental CHO-K1 cells or CD123-transfected CHO-K1 cells. The dilutions of 7G3 and 6H6 were also used as a comparison. As shown in Figure 8A,B, when coupled with DTG3, 12F1 and 12H12 had an IC50 of approximately 42 pM, very similar to what was found with initial culture supernatants from 96-well plates (Table 1). These antibodies alone had no effect on cell growth (data not shown). 12F1 and 12H12 had ∼2-10-fold higher internalization activities than existing antibodies and appeared to be the best candidates for immunotoxin production.

DISCUSSION Current monoclonal antibody production utilized in immunotoxin design is based on the selection of hybridoma cell lines that are able to produce large quantities of antibodies that have high affinity for the intended target. However, this approach

Kuo et al.

Figure 8. Cytotoxicity assays. Anti-CD123 antibodies-DTG3 complexes were tested in (A) parental CHO-K1 cells and (B) CD123-transfected CHO-K1 cells for their potencies.

may not be adequate for the design of successful conjugates, as immunotoxin efficacy is critically dependent on antibody internalization after cell surface antigen binding. We have reported here the development of DTG3, a novel protein conjugate specifically for the quantification of antibody internalization. Conventional chemical conjugation of monoclonal antibodies to the cell-surface binding deficient mutant diphtheria toxin CRM9 was previously used to determine antibody internalization efficiency. However, this method has proven too costly, inefficient, and time-consuming. In order to minimize wasted reagents, the reaction was scaled down by our laboratory. Regardless, 10 mg CRM9 and 2 mg antibody were still required for the acquisition of adequate amounts of immunoconjugates for cytotoxicity analysis. Purification of the conjugate from free toxin and unconjugated antibody also proved to be timeconsuming and tedious. After product purification and concentration determination, enough pure products remained for 1-2 cytotoxicity assays. Further, since chemical linkers are randomly conjugated to nonspecific sites on the proteins, activity of functional groups has the potential to be reduced, causing false negative results and low reproducibility between batches. Conversely, DTG3 fusion proteins require only 2 to 3 µg of antibody per duplicate cytotoxicity assay. The preferential binding of Protein G to the immunoglobulin Fc region permits full exposure of the antigen binding domains. In addition, this system has proven to be extremely time efficient. Coupling reactions can be performed in about 1 h without any additional purification or concentration of the final product. Therefore,

Antibody Internalization Assays

these results indicate that the utilization of DTG3 facilitates the investigation of the internalization efficiency of monoclonal antibodies, and thus eases the cost and labor of immunotoxin development. As shown in Figure 6 and Table 1, we are able to use culture supernatant at the earliest stage of screening to examine their ability to be internalized upon binding to the cell surface antigens. Inspiration for the generation of this novel DTG3 fusion protein was achieved from earlier work by Mazor et al. (8). This group demonstrated that a fusion protein consisting of the Pseudomonas exotoxin (PE) and two copies of Protein A IgG binding domain, known as ZZ-PE38, was useful to determine antibody internalization efficiency. However, several considerations resulted in numerous modifications in order to optimize this system for the application in our laboratory. Primarily, a specific cellular trafficking mechanism is required for PE to reach the cytosol, inhibit ribosome function, and thus kill the target cells. It is known that many cell types, including myeloid cells, have an inefficient translocation mechanism and are more resistant to PE (29). Our primary interests in myeloid cell malignancies prompted the substitution of PE to DT in order to circumvent this faulty trafficking mechanism, since the catalytic subunit of diphtheria toxin (DT) can be released from endosome to cytosol directly (30, 31). Additionally, Protein A demonstrates a differential affinity for immunoglobulins based on the species of origin and immunoglobulin iso-types. It has been shown that murine monoclonal antibody iso-type IgG2 has the highest affinity for Protein A, causing a limited usefulness of ZZ-PE38. However, Protein G has a much broader spectrum of immunoglobulin binding ability. Protein G is therefore chosen in order to enhance the convenience of this assay. Finally, our studies indicate that DTG3 was able to bind all antibodies tested within 60 min at a 1:1 ratio. In contrast to the published data (8), further purification of antibody-DTG3 complexes is not necessary. However, titration of the antibodyDTG3 complexes into cell culture medium without prestabilization by DMP will result in the dissociation of these complexes via exchange reactions with immunoglobulins in the media serum (data not shown). In order to eliminate the competition effects, we adapted Jurkat cells to serum free AIM-V medium (Invitrogen) and used them in UCHT1-DTG3 potency assays. Indeed, under serum free condition, UCHT1-DTG3 without pretreated with DMP had an IC50 in Jurkat cells at ∼40 pM (data not shown). However, this number is still about 2 orders higher than DMP-stabilized UCHT1-DTG3 complexes (Figure 3). After consulting with technical service at Invitrogen, they indicated that the BSA added in the AIM-V medium could contain a low amount of immunoglobulin. When UCHT1-DTG3 was titrated down to ∼10 pM or 1 ng/mL, a small amount of immunoglobulin contaminant could have a big effect. These findings indicate that the inclusion of DMP to stabilize antibodyDTG3 complexes is essential for the success of this assay. In conclusion, antibody internalization by the target cell is critical in immunotoxin development. We have described a rapid protocol to quantify antibody internalization via potency on the target cell. The development and validation of this user-friendly system has potential to facilitate the design of viable immunotoxins and thus reduce the time and cost needed for clinical development.

ACKNOWLEDGMENT We would like to thank Drs. JH Woo, NH Bander, DM Neville, and M Sadelain for providing antibodies and cell lines, Johnny Lee for assisting in the operation of flow cytometer, and Kent Claypool (Institute for Regenerative Medicine, TAMU) for performing cell sorting to isolate transfected CHO-

Bioconjugate Chem., Vol. 20, No. 10, 2009 1981

K1 cells with high copy number of cell surface CD123. This project is supported by Scott & White setup fund to JSL.

LITERATURE CITED (1) Pastan, I., Hassan, R., FitzGerald, D. J., and Kreitman, R. J. (2007) Immunotoxin treatment of cancer. Annu. ReV. Med. 58, 221–37. (2) Reichert, J. M., and Valge-Archer, V. E. (2007) Development trends for monoclonal antibody cancer therapeutics. Nat. ReV. Drug DiscoVery 6, 349–56. (3) Jordan, C. T., Upchurch, D., Szilvassy, S. J., Guzman, M. L., Howard, D. S., Pettigrew, A. L., Meyerrose, T., Rossi, R., Grimes, B., Rizzieri, D. A., Luger, S. M., and Phillips, G. L. (2000) The interleukin-3 receptor alpha chain is a unique marker for human acute myelogenous leukemia stem cells. Leukemia 14, 1777–84. (4) Sun, Q., Woodcock, J. M., Rapoport, A., Stomski, F. C., Korpelainen, E. I., Bagley, C. J., Goodall, G. J., Smith, W. B., Gamble, J. R., Vadas, M. A., and Lopez, A. F. (1996) Monoclonal antibody 7G3 recognizes the N-terminal domain of the human interleukin-3 (IL-3) receptor alpha-chain and functions as a specific IL-3 receptor antagonist. Blood 87, 83–92. (5) Rappoport, J. Z. (2008) Focusing on clathrin-mediated endocytosis. Biochem. J. 412, 415–23. (6) Rapoport, A. P., Luhowskyj, S., Doshi, P., and DiPersio, J. F. (1996) Mutational analysis of the alpha subunit of the human interleukin-3 receptor. Blood 87, 112–22. (7) Zangemeister-Wittke, U. (2005) Antibodies for targeted cancer therapy -- technical aspects and clinical perspectives. Pathobiology 72, 279–86. (8) Mazor, Y., Barnea, I., Keydar, I., and Benhar, I. (2007) Antibody internalization studied using a novel IgG binding toxin fusion. J. Immunol. Methods 321, 41–59. (9) Frankel, A., Liu, J. S., Rizzieri, D., and Hogge, D. (2008) Phase I clinical study of diphtheria toxin-interleukin 3 fusion protein in patients with acute myeloid leukemia and myelodysplasia. Leuk. Lymphoma 49, 543–53. (10) Black, J. H., McCubrey, J. A., Willingham, M. C., Ramage, J., Hogge, D. E., and Frankel, A. E. (2003) Diphtheria toxininterleukin-3 fusion protein (DT388IL3) prolongs disease-free survival of leukemic immunocompromised mice. Leukemia 17, 155–9. (11) Cohen, K. A., Liu, T. F., Cline, J. M., Wagner, J. D., Hall, P. D., and Frankel, A. E. (2004) Toxicology and pharmacokinetics of DT388IL3, a fusion toxin consisting of a truncated diphtheria toxin (DT388) linked to human interleukin 3 (IL3), in cynomolgus monkeys. Leuk. Lymphoma 45, 1647–56. (12) Urieto, J. O., Liu, T., Black, J. H., Cohen, K. A., Hall, P. D., Willingham, M. C., Pennell, L. K., Hogge, D. E., Kreitman, R. J., and Frankel, A. E. (2004) Expression and purification of the recombinant diphtheria fusion toxin DT388IL3 for phase I clinical trials. Protein Expr. Purif. 33, 123–33. (13) Cohen, K. A., Liu, T. F., Cline, J. M., Wagner, J. D., Hall, P. D., and Frankel, A. E. (2005) Safety evaluation of DT388IL3, a diphtheria toxin/interleukin 3 fusion protein, in the cynomolgus monkey. Cancer Immunol. Immunother. 54, 799–806. (14) Garnache-Ottou, F., Feuillard, J., and Saas, P. (2007) Plasmacytoid dendritic cell leukaemia/lymphoma: towards a well defined entity? Br. J. Hamaetol. 136, 539–48. (15) Djokic, M., Bjorklund, E., Blennow, E., Mazur, J., Soderhall, S., and Porwit, A. (2009) Overexpression of CD123 correlates with the hyperdiploid genotype in acute lymphoblastic leukemia. Haematologica 94, 1016–9. (16) Orazi, A., Chiu, R., O’Malley, D. P., Czader, M., Allen, S. L., An, C., and Vance, G. H. (2006) Chronic myelomonocytic leukemia: The role of bone marrow biopsy immunohistology. Mod. Pathol. 19, 1536–45. (17) Florian, S., Sonneck, K., Hauswirth, A. W., Krauth, M. T., Schernthaner, G. H., Sperr, W. R., and Valent, P. (2006) Detection of molecular targets on the surface of CD34+/CD38--

1982 Bioconjugate Chem., Vol. 20, No. 10, 2009 stem cells in various myeloid malignancies. Leuk. Lymphoma 47, 207–22. (18) Riccioni, R., Diverio, D., Riti, V., Buffolino, S., Mariani, G., Boe, A., Cedrone, M., Ottone, T., Foa, R., and Testa, U. (2009) Interleukin (IL)-3/granulocyte macrophage-colony stimulating factor/IL-5 receptor alpha and beta chains are preferentially expressed in acute myeloid leukaemias with mutated FMS-related tyrosine kinase 3 receptor. Br. J. Hamaetol. 144, 376–87. (19) Du, X., Ho, M., and Pastan, I. (2007) New immunotoxins targeting CD123, a stem cell antigen on acute myeloid leukemia cells. J. Immunother. 30, 607–13. (20) Liu, T. F., Urieto, J. O., Moore, J. E., Miller, M. S., Lowe, A. C., Thorburn, A., and Frankel, A. E. (2004) Diphtheria toxin fused to variant interleukin-3 provides enhanced binding to the interleukin-3 receptor and more potent leukemia cell cytotoxicity. Exp. Hematol. 32, 277–81. (21) Sjobring, U., Bjorck, L., and Kastern, W. (1991) Streptococcal protein G. Gene structure and protein binding properties. J. Biol. Chem. 266, 399–405. (22) Frankel, A. E., Ramage, J., Kiser, M., Alexander, R., Kucera, G., and Miller, M. S. (2000) Characterization of diphtheria fusion proteins targeted to the human interleukin-3 receptor. Protein Eng. 13, 575–81. (23) Kuo, S. R., Liu, J. S., Broker, T. R., and Chow, L. T. (1994) Cell-free replication of the human papillomavirus DNA with homologous viral E1 and E2 proteins and human cell extracts. J. Biol. Chem. 269, 24058–65. (24) Kiser, M., McCubrey, J. A., Steelman, L. S., Shelton, J. G., Ramage, J., Alexander, R. L., Kucera, G. L., Pettenati, M., Willingham, M. C., Miller, M. S., and Frankel, A. E. (2001) Oncogene-dependent engraftment of human myeloid leukemia cells in immunosuppressed mice. Leukemia 15, 814–8.

Kuo et al. (25) Yokoyama, W. M. Christensen, M., Santos, G. D., and Miller, D. (2006) Production of monoclonal antibodies, In Current Protocols in Immunology, Chapter 2, Unit 2.5, John Wiley & Sons, New Jersey. (26) Davis, J. M. (1986) A single-step technique for selecting and cloning hybridomas for monoclonal antibody production. Methods Enzymol. 121, 307–22. (27) Woo, J. H., Liu, J.-S., Kang, S. H., Singh, R., Park, S. K., Su, Y., Ortiz, J., Neville, D. M., Jr., Willingham, M. C., and Frankel, A. E. (2008) GMP production and characterization of the bivalent anti-human T cell immunotoxin, A-dmDT390bisFv(UCHT1) for phase I/II clinical trials. Protein Expr. Purif. 58, 1–11. (28) Liu, H., Rajasekaran, A. K., Moy, P., Xia, Y., Kim, S., Navarro, V., Rahmati, R., and Bander, N. H. (1998) Constitutive and antibody-induced internalization of prostate-specific membrane antigen. Cancer Res. 58, 4055–60. (29) Frankel, A. E., Hall, P. D., Burbage, C., Vesely, J., Willingham, M., Bhalla, K., and Kreitman, R. J. (1997) Modulation of the apoptotic response of human myeloid leukemia cells to a diphtheria toxin granulocyte-macrophage colony-stimulating factor fusion protein. Blood 90, 3654–61. (30) Falnes, P. O., and Sandvig, K. (2000) Penetration of protein toxins into cells. Curr. Opin. Cell Biol. 12, 407–13. (31) Collier, R. J. (2001) Understanding the mode of action of diphtheria toxin: a perspective on progress during the 20th century. Toxicon 39, 1793–803. (32) Grussenmeyer, T., Scheidtmann, K. H., Hutchinson, M. A., Eckhart, W., and Walter, G. (1985) Complexes of polyoma virus medium T antigen and cellular proteins. Proc. Natl. Acad. Sci. U.S.A. 82, 7952–4. BC900333J