|Home | About | Journals | Submit | Contact Us | Français|
The selective cell surface expression of receptor tyrosine kinase-like orphan receptor 1 (ROR1) in chronic lymphocytic leukemia (CLL) and mantle cell lymphoma (MCL) has made ROR1 a novel and promising target for therapeutic monoclonal antibodies (mAbs). Four mouse mAbs generated by hybridoma technology exhibited specific binding to human ROR1. Epitope mapping studies showed that two mAbs (2A2 and 2D11) recognized N-terminal epitopes in the extracellular region of ROR1 and the other two (1A1 and 1A7) recognized C-terminal epitopes. A ROR1- immunotoxin (BT-1) consisting of truncated Pseudomonas exotoxin A (PE38) and the VH and VL fragments of 2A2-IgG was made recombinantly. Both 2A2-IgG and BT-1 showed dose-dependent and selective binding to primary CLL and MCL cells and MCL cell lines. Kinetic analyses revealed 0.12-nM (2A2-IgG) to 65-nM (BT-1) avidity/affinity to hROR1, depicting bivalent and monovalent interactions, respectively. After binding to cell surface ROR1, 2A2-IgG and BT-1 were partially internalized by primary CLL cells and MCL cell lines, and BT-1 induced profound apoptosis of ROR1-expressing MCL cell lines in vitro (EC50 = 16 pM–16 nM), but did not affect ROR1-negative cell lines. Our data suggest that ROR1-immunotoxins such as BT-1 could serve as targeted therapeutic agents for ROR1-expressing B cell malignancies and other cancers.
Receptor tyrosine kinases (RTKs) are known to be key regulators of normal cellular processes such as differentiation, migration, proliferation and survival, but they also have a critical role in the development and progression of many types of human cancers. Consequently, RTKs and their ligands have become attractive molecular targets for therapeutic interventions of cancer. The ROR family of RTKs consists of two evolutionarily conserved proteins, ROR1 and ROR21 and gene expression profiling identified ROR1 as one of the signature genes overexpressed in CLL.2,3 Subsequent reports showed a variety of cancer cell lines and primary cancer cells of different histological origin expressed significant levels of ROR1 mRNA.4,5 Screening of the human genome revealed ROR1 as one of the few potent survival kinases,6 and siRNA-mediated silencing of ROR1 induced apoptosis in CLL cells.7 ROR1 protein levels increased following IL-6-induced phosphorylation of a transcription factor, signal transducer and activator of transcription 3 (STAT-3), in myeloma cells, and overexpression of STAT-3, multiple Wnt genes and Wnt receptor Frizzled was observed in CLL cells.8 Unlike other RTKs, ROR1 is catalytically inactive and is transphosphorylated by the Met oncogene.5 Together, these findings suggest that ROR1 could provide survival signals and may contribute in the development of leukemia. Studies from our laboratory and others have demonstrated the uniform and restricted cell surface expression of ROR1 protein in CLL.9-11 We therefore hypothesized that ROR1 could serve as a target antigen for mAb based therapeutics.
B cell malignancies are susceptible to therapeutic intervention by mAbs, and rituximab and ofatumumab, both of which target CD20 on all B cells (malignant and normal), and alemtuzumab, which targets CD52 on all leukocytes (malignant and normal), are FDA-approved for B cell non-Hodgkin lymphoma (B-NHL) or CLL. Mechanistically, these three unarmed mAbs mediate direct cytotoxicity, complement dependent cytotoxicity (CDC) or antibody dependent cellular cytotoxicity (ADCC), and their efficacy depends on the availability of functional effector cells/molecules in the recipient. Although treatment with these mAbs has achieved significant progression-free survival and short-term overall survival benefits, most patients relapse or develop resistance. Therefore, there is a need for identification of new targets and development of novel therapeutics. In this context, mAbs against ROR1 would be welcome as they could specifically target malignant B cells, thereby sparing normal B cells and other leukocytes. Interestingly, patients with CLL produced endogenous anti-ROR1 antibodies following vaccination with CD40L-expressing autologous tumor cells,10 and following administration of the immunomodulatory drug lenalidomide,12 which upregulated CD40L expression in vitro. These observations suggest that antibodies directed against ROR1 per se may not cause any adverse events in vivo, although their contribution to alleviate the disease remains to be understood. The restricted expression of cell surface ROR1 on malignant B cells and the apparent safety of an endogenous anti-ROR1 antibody response, prompted us to develop a ROR1-immunotoxin. Unlike unarmed mAbs, a ROR1-immunotoxin could specifically recognize and kill the target cells independent of effector components (complement proteins, NK cells, macrophages or T cells) that may not be optimally available in cancer patients. Immunotoxins are fusion proteins consisting of a target antigen-binding moiety (a ligand or mAb) that confers specificity, and a toxin moiety (from a plant or bacterium) that mediates target cell killing.13,14 Chemical conjugation of mAb and toxin has resulted in heterogeneous products unsuitable for clinical applications; this problem was overcome by genetic engineering wherein gene(s) encoding an antibody fragment are linked to a gene encoding selected toxin domains.14-18 A variety of immunotoxins with different antigen target specificities and different toxins have been developed to treat human cancers.19-22 We used truncated pseudomonas exotoxin A (PE38) because the functional domains of this toxin have been well characterized and the mechanisms of killing are known.16,23 Several PE38-based immunotoxins have been developed and are currently in clinical trials for different human disease indications, including B cell malignancies.20,24-27 We describe here the generation and functional characterization of the PE38-based immunotoxin BT-1 directed against ROR1.
Patients with CLL and MCL were enrolled in institutional review board–approved studies at the National Institutes of Health (NIH), Bethesda, MD and blood samples were obtained with written informed consent. Blood samples from healthy donors were obtained from the Department of Transfusion Medicine, Clinical Center, NIH. Peripheral blood mononuclear cells (PBMC) were prepared by density gradient separation of whole blood on lymphocyte separation medium (MP Biochemicals, #50494) and cryopreserved until use. The CLL cell lines MEC-1 and MEC-2 were obtained from DSMZ (Braunschweig, Germany) and 232-B4 was a gift from Dr. Anders Rosén (Linköping University, Linköping, Sweden). The MCL cell lines used in this study (JeKo-1, HBL-2, Mino and Rec-1) have been characterized,28 and Burkitt lymphoma cell lines (Raji, Ramos and Daudi) were obtained from ATCC (Manassas, VA). EBV-transformed lymphoblastoid cell lines were generated from PBMC from a healthy donor (EBV-7475) and a patient with CLL (RGC-EBV) by following standard procedure;29 both of these were oligoclonal suggesting their origin from normal B cells (unpublished observation). All cell lines were maintained in vitro in RPMI-1640 supplemented with 2 mM Glutamax, 10% (v/v) (GIBCO #61870) fetal bovine serum (FBS) (GIBCO #10437), 20 mM HEPES (GIBCO #15630080), 100 IU/mL penicillin, 100 μg/mL streptomycin sulfate (GIBCO #15140–148), 50 μM β-mercaptoethanol (Sigma #M-7522) and 1 mM sodium pyruvate (GIBCO #11360070).
A panel of recombinant ROR1 proteins were cloned, expressed and purified as described elsewhere:30 Fc-hROR1, hROR1-ECD, Fc-hROR1Ig+Fz (Ig+Fz), Fc-hROR1Fz+Kr (Fz+Kr), Fc-hROR1Ig (Ig), Fc-hROR1Fz (Fz), Fc-hROR1Kr (Kr), human Fc (Fc) and Fc-mROR1. Purified human ROR2-Fc was purchased from R&D Systems (Custom-MO2).
Mouse anti-hROR1 mAbs were custom generated (Precision Antibody, Columbia, MD). In brief, mice were immunized with Fc-hROR1, and after three booster injections with hROR1-ECD serum antibody response against ROR1 was detected. Hybridomas were generated using the immune spleen cells. Screening of about 100 hybridoma clones yielded four clones 2A2, 2D11, 1A1 and 1A7 that secreted mAbs that bound specifically to ROR1. The four hybridoma clones were grown in BD Cell mAb medium (Animal Component Free, Becton Dickinson #220513) using CELLine Flasks (Becton Dickinson #353137) and supernatants were collected. HiTrap Protein G columns (GE Healthcare #17–0404–01) were used to purify immunoglobulins (Ig) from the supernatants. Reduced SDS-PAGE analysis revealed only Ig heavy and light chain bands of expected molecular sizes for each mAb. The isotype of all four mAbs was determined to be mouse IgG1-kappa using the Mouse Immunoglobulin Isotyping ELISA Kit (BD PharMingen #550487).
Partial amino acid sequences from both variable domains of 2A2-IgG were obtained by custom LC-MS/MS analysis on tryptic digests of SDS-PAGE resolved heavy and light chain polypeptides (Alphalyse, Custom order). Based on this information, the mouse V gene families were identified using Ig BLAST (www.ncbi.nlm.nih.gov/igblast). Oligonucleotides aligning to the signal peptide encoding DNA sequences of the Vκ and VH gene families and to the Cκ and Cγ11 constant domain encoding DNA sequences were used to amplify 2A2-VH and 2A2-VL by RT-PCR from total RNA isolated from 2A2 hybridoma cells. DNA sequencing of the PCR products yielded the complete amino acid sequences of the 2A2-VH and 2A2-VL polypeptides (International Patent Application PCT/US2010/032208). The 2A2-based immunotoxin BT-1 was generated following the procedure described elsewhere with some changes detailed below.31 Optimized for expression in Escherichia coli (E.coli), the DNA sequences encoding 2A2-VH and 2A2-VL polypeptides were custom synthesized with the following modifications (DNA 2.0, Menlo Park, CA); 2A2-VH included a NdeI site at the 5′ end and a HindIII site at the 3′ end; 2A2-VL included a NdeI site at the 5′ end and an EcoRI site at the 3′ end. In addition, the codons for residues 45 in 2A2-VH and 99 in 2A2-VL, both present in framework regions, were substituted with a codon for cysteine to allow pairing of VH and VL polypeptides and yield disulphide linked Fv (dsFv) after refolding. These DNA sequences were cloned into separate stocks of pJ201 plasmids. NdeI/HindIII-digested 2A2-VH was cloned into a T7 expression vector (pRB98) creating an in-frame fusion with a DNA sequence encoding truncated Pseudomonas exotoxin A (PE38). Separately, NdeI/EcoRI-digested 2A2-VL was cloned into pRB98 (thereby removing the PE38 encoding DNA sequence).31 The two expression plasmids were separately electroporated into E. coli strain BL21 (λDE3) cells (ElectroMax; Invitrogen #11319–019) and 1 L culture of each plasmid was grown with IPTG induction. The bacterial pellets were processed and inclusion body proteins were pooled and allowed to refold. The refolded ROR1-immunotoxin, 2A2(dsFv)-PE38 (named BT-1), was purified by two sequential ion exchange chromatography steps using Q-Sepharose column (GE Healthcare #17–5159–01) and Mono-Q column (GE Healthcare #17–5166–01) followed by a gel filtration chromatography step using a HiLoad 16/60 Superdex S200 prep grade column (GE Healthcare #17–1069–01), all in conjunction with an ÄKTA-FPLC system using Unicorn software (Amersham Pharmacia Biotech, Sweden). The purity of BT-1 was validated by a single sharp peak seen in the gel filtration chromatography profile, its retention time was in close proximity with that of bovine serum albumin (BSA, Sigma-Aldrich #A-7906) in the same column, and SDS-PAGE analysis revealed a single band of the expected size (Supplementary Figure 2S). Purified BT-1 was stored in PBS pH 7.2 at -70°C until use.
To determine the specific reactivity and epitope mapping of anti-ROR1 mAbs, 96-well plates (Clear Microplate; R&D Systems #DY990) were coated with indicated recombinant ROR1 proteins (100 ng in 50 μL/well) in PBS pH 7.2 overnight at 4°C. The next day, the wells were washed with 10 mM Tris-buffered saline with 0.1% (v/v) Triton X-100 pH 7.2 (TBST) using an automatic plate washer (Beckman Coulter, Brea, CA), and blocked with 300 μL/well 3% (w/v) BSA in PBS pH 7.2 for 2 h at 25°C. After one wash cycle, 50 μL of primary anti-ROR1 antibodies (mouse mAbs 2A2-IgG, 2D11-IgG, 1A1-IgG, 1A7-IgG, goat polyclonal antibodies (pAbs) against ROR1 (R&D Systems #AF2000), goat pAbs against ROR2 (R&D Systems #AF2064) or BT-1) diluted in 1% (w/v) BSA in PBS pH 7.2 were added in different concentrations and incubated at 25°C for 2 h. For BT-1 alone, the wells were washed again, 50 μL/well polyclonal rabbit anti-pseudomonas endotoxin A serum (RaPEA, Sigma-Aldrich #P2318) was added at 1:300 final dilution and incubated for 2 h. After two wash cycles, 50 μL/well corresponding detection antibodies were added (horseradish peroxidase conjugated donkey anti-mouse IgG, DaM-HRP (Jackson ImmunoResearch Laboratories #715–036–150); donkey anti-goat IgG, DaG-HRP (#705–035–147) or donkey anti-rabbit IgG, DaR-HRP (#711–036–152) at 1:1,000 dilution in 1% (w/v) BSA in PBS pH 7.2 and incubated for 1 h at 25°C. After two wash cycles, the enzymatic activity of bound antibody was determined by adding the HRP substrate ABTS (Roche Applied Science #10102946001) and measuring the absorbance at 405 nm (and 490 nm to allow auto correction) in a VersaMax microplate reader (Molecular Devices, Sunnyvale, CA).
The binding kinetics of anti-hROR1 antibody formats (2A2-IgG and BT-1) and hROR1 were studied by measuring the surface plasmon resonance using a Biacore X-100 instrument and Biacore reagents and software (GE Healthcare). This was done either by direct coupling or capture of the ligands at comparable and low densities to minimize the mass transfer effect. For direct coupling, a CM5 sensor chip (GE Healthcare #BR-1000–12) was activated with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride and N-hydroxysuccinimide for subsequent ligand immobilization. The fusion proteins Fc-hROR1 and hROR1-ECD in 10 mM sodium acetate pH 5.0 were immobilized in flow cell 1 (FC-1) at a density of 160–300 resonance units (RU) in separate sensor chips. The FC-2 in each chip served as control (without ligand) to allow instant background correction. The chips were deactivated with 1 M ethanolamine hydrochloride, pH 8.5. The analytes, 2A2-IgG and BT-1, were diluted in 1x HBS-EP+ running buffer pH 7.4 (10 mM HEPES, 150 mM NaCl, 3 mM EDTA and 0.05% (v/v) Surfactant P20), and five different concentrations of analytes (range 2.5–100 nM) were run in duplicates at a flow rate of 30 μL/min. The chips were regenerated using glycine-HCl pH 2.0 without significant loss of binding capacity. In the capture method, first polyclonal rabbit anti-mouse IgG was immobilized on a CM5 chip to give an approximate final density of 9790 RU in both FC-1 and FC-2 (Mouse Antibody Capture Kit; GE Healthcare #BR-1008–38) and deactivated following the manufacturer’s recommended procedure. Subsequently, the ligand (2A2-IgG) was captured at 1 μg/mL (predetermined concentration) to give an approximate density of 200 RU in FC-1, with FC-2 serving as control. The analytes (hROR1-ECD and Fc-hROR1) were run at five different concentrations (range 3–1,000 nM) in duplicates as described above. Calculation of association (kon) and dissociation (koff) rate constants was based on a 1:1 Langmuir binding model for both monovalent and bivalent interactions. The equilibrium dissociation constant (KD) was calculated from koff /kon.
Approximately 5 x 105 cells were incubated with different concentrations of affinity-purified 2A2-IgG or normal polyclonal mouse IgG (mIgG; Jackson ImmunoResearch Laboratories # 015–000–003) for 45 min at 4°C. After two washes with FACS buffer (2% (v/v) FBS in PBS pH 7.2), the cells were incubated for 45 min at 4°C with DyLight-649 conjugated, affinity pure F(ab)2 preparation of goat anti-mouse IgG (Dy-GaM; Jackson ImmunoResearch Laboratories #115–496–146) or APC-conjugated goat anti-mouse IgG, Fc fragment specific (APC-GaM; # 115–136–071) at 1:500 final dilution. For primary CLL and MCL cells, the malignant B cells were gated using CD5/CD19 Simultest reagent (BD Biosciences #340396). Similarly, for staining with ROR1-immunotoxin BT-1, target cells were incubated with various concentrations of BT-1, followed by normal polyclonal rabbit IgG (rIgG, Jackson ImmunoResearch Laboratories #011–000–002) or the secondary antibody (RaPEA) at 1:300 final dilution and then with DyLight-649-conjugated, affinity pure F(ab)2 preparation of goat anti-rabbit IgG (Dy-GaR; Jackson ImmunoResearch Laboratories #111–496–144) at 1:500 final dilution. Cells were washed twice with FACS buffer in each step and, after two final washes, they were resuspended in FACS buffer and 5 μg/mL propidium iodide (P.I. Sigma-Aldrich #P4170) was added to exclude dead cells from analysis. For the detection of intracellular BCL2, the cells were washed in PBS, then with BD Perm/Wash buffer (BD Biosciences #554723), and subsequently fixed and permeabilized for 20 min using BD Cytofix/Cytoperm buffer (BD Biosciences #554722) and washed again with BD Perm/Wash buffer. The cells were stained with PE-conjugated mouse anti-human BCL2 mAb or an isotype control (BD Biosciences, BCL2 set #556535). A total of 20000 gated events were collected for each sample using FACSCalibur instrument (BD Biosciences) and data were analyzed using CellQuest software. The values were depicted as mean fluorescence intensity (MFI).
ROR1-mediated internalization was determined by measuring the decrease in surface bound (at 4°C) anti-ROR1 mAb after incubation at 37°C.9 Approximately 2 x 106 target cells were incubated with 1 μg/mL unconjugated 2A2-IgG for 1 h on ice (4°C) in RPMI-5% (v/v) FBS. The cells were washed twice with ten volumes of medium and resuspended as 5 x 105 aliquots in medium. One aliquot was left on ice and the others were incubated at 37°C for various periods of time either without or with 10 μM phenylarsine oxide (PAO; Sigma-Aldrich #P3075). Next, the cells were washed with medium and all samples were incubated with Dy-GaM at 1:500 final dilution. After 1 h incubation on ice, the cells were washed and resuspended in FACS buffer (see above) with 5 μg/mL P.I. and analyzed by flow cytometry. Additional experiments were performed using biotinylated 2A2-IgG prepared with the BiotinTag Microbiotinylation kit (Sigma-Aldrich #BIOTAG), and detected by streptavidin-conjugated phycoerythrin (SA-PE, BD Biosciences #554061). Furthermore, directly labeled 2A2-IgG was used to simultaneously measure cell surface bound and internalized antibody. For this, mIgG (as control) and 2A2-IgG were coupled to Atto-488 following the manufacturer’s recommended procedure (Novus Biologicals 733–0010). While Atto488 fluorescence collectively measured cell surface bound and internalized 2A2-IgG, only the cell surface bound fraction was detectable with a secondary antibody (Dy-GaM) in non-permeabilized cells. Internalization of BT-1 was analyzed by sequential incubations with RaPEA and Dy-GaR (Jackson ImmunoResearch Laboratories # 111–496–144) as described above.
The number of ROR1 molecules on the target cell surface was determined using the Quantum Simply Cellular Microbeads Kit (Bangs Laboratories #815B) as described before,9 with following modifications. First, 2A2-IgG was directly labeled using Alexa Fluor 647 Protein Labeling Kit (Molecular Probes #A20173) following the recommended procedure to obtain 2A2-IgG-AF647 conjugate. In parallel, polyclonal mouse IgG (mIgG) was used to obtain mIgG-AF647. Approximately, 1 x 106 cells were incubated for 30 min at 4°C in the dark with predetermined saturating concentration of 2A2-IgG-AF647. As control, mIgG-AF647 was used. In parallel, the quantum beads were incubated with predetermined saturating concentration of 2A2-IgG-AF647 conjugate. The cells and the beads were washed three times, resuspended in the FACS buffer and the MFI were determined by flow cytometry with the same instrument settings. Using the QuickCal software (Bangs Laboratories), a calibration curve was generated by plotting MFI vs. the antigen binding capacity (ABC) of the beads. The delta MFI values of each sample (2A2-IgG-AF647 minus mIgG-AF647) was used to determine the ABC of the sample and considered to represent the antigen (ROR1) density on the cell surface.
ROR1-immunotoxin (BT-1) induced death of target cells was detected by three independent methods: a) detection of cell surface exposure of phosphatidylserine (PS),32 b) measurement of changes in mitochondrial membrane potential33 and c) intracellular detection of activated caspase-3.34,35 ROR1-expressing and non-expressing cell lines (2.5 x 105 cells) were incubated at 37°C for various periods of time in RPMI-5% (v/v) FBS in the absence or presence of different amounts of BT-1 (0.01 – 100 μg/mL). For PS exposure, both BT-1 treated and untreated cells were washed in Annexin-V binding buffer (ABB), incubated with Alexa Fluor 647-conjugated Annexin-V (Invitrogen #A23204) and SYTOX Green (Invitrogen #S7020) following the manufacturer’s recommended procedure and subjected to flow cytometry. Camptothecin (Sigma-Aldrich #C9911) was used (10 μM) as a positive control to induce apoptosis. The percentage of apoptotic cells was determined as follows: % Apoptosis = % Annexin-V-positive cells + % Annexin-V-positive and SYTOX Green-positive cells/total cells analyzed x 100. The SYTOX Green-positive and Annexin-V-negative cells were considered necrotic cells, and the double negative cells were considered live cells. The change in mitochondrial membrane potential (ΔmΨ) has been considered to be one of the early events in apoptosis, and such a change leads to accumulation of cationic cyanine dyes in cells that can be measured by flow cytometry. Untreated and BT-1 treated cells were incubated with a predetermined concentration (12.5 nM) of 3,3′-diethyloxacarbocyanine iodide (DiOC2(3); MitoProbe kit, Molecular Probes #M34150) following the manufacturer’s recommended procedure. The cells were washed once and red and green fluorescence intensities were measured. Red fluorescence intensity increases due to dye stacking in cells with active mitochondria and decreases in cells with disrupted mitchondria. Both qualitative and quantitative changes in fluorescence intensity of the dye were measured and depicted as histogram. The caspase family of cysteine proteases plays a key role in apoptosis. In particular, caspase-3 is activated during early stages of apoptosis, and the active enzyme cleaves survival proteins such as BCL2 and PARP, resulting in apoptosis. To determine intracellular levels of activated caspase-3, BT-1 treated and untreated cells were washed in PBS, fixed and permeabilized using BD Cytofix/Cytoperm for 20 min at 25°C, washed with BD Perm/Wash buffer. Cells were then stained with PE-conjugated rabbit anti-active caspase-3 antibody (BD Biosciences #550821) and subjected to flow cytometry.
The specificity of mAbs (2A2-IgG, 2D11-IgG 1A1-IgG and 1A7-IgG), and the epitopes they recognized were determined by studying their reactivity against a panel of recombinant fusion proteins using ELISA (Fig. 1). All four mAbs exhibited strong binding to Fc-hROR1 and hROR1-ECD in a concentration-dependent manner (range tested 10 μg/mL to 1.5 ng/mL, data shown for only 1 μg/mL), but not to human Fc, hROR2-Fc or BSA indicating their specific reactivity to purified hROR1. In addition, while both 2A2-IgG and 2D11-IgG were able to bind Fc-hROR1Ig+Fz, only 2A2-IgG bound to Fc-hROR1Ig and neither mAb bound to Fc-hROR1Fz, Fc-hROR1Kr or Fc-hROR1Fz+Kr proteins. This pattern of reactivity suggested that the epitope recognized by mAb 2A2 is likely located in the Ig domain of hROR1and that of 2D11 mAb is shared between Ig and Fz domains. Conversely, both 1A1-IgG and 1A7-IgG were able to bind Fc-hROR1Fz+Kr and Fc-hROR1Kr, but not to Fc-hROR1Ig+Fz, Fc-hROR1Ig or Fc-hROR1Fz proteins, suggesting that the epitopes recognized by these two mAbs are likely located in the Kr domain of hROR1. As positive controls, GaROR1 pAbs demonstrated specific binding to all recombinant hROR1 proteins and GaROR2 pAbs to the recombinant hROR2 protein (data not shown). In addition, the mAbs exhibited moderate (2D11-IgG, 1A7-IgG) to strong (2A2-IgG, 1A1-IgG) cross-reactivity to mouse ROR1 (Fc-mROR1). Overall, mAb 2A2 exhibited stronger reactivity to hROR1 and mROR1 than the other three mAbs, which was confirmed by surface plasmon resonance studies (see below and data not shown). This prompted us to focus on mAb 2A2 and clone, express and purify a mAb 2A2-based immunotoxin in dsFv format we named BT-1 (see Materials and Methods and Supplementary Figure 2S). Like the parent mAb 2A2, BT-1 showed specific binding to hROR1-ECD in ELISA (OD 405 value = 1.38) and did not bind hROR2-Fc (OD 405 value = 0, i.e., equal to background).
Both 2A2-IgG and 2D11-IgG exhibited dose-dependent (0.001 – 10 µg/mL) binding to MCL cell lines JeKo-1, Mino and ex vivo primary CLL and MCL cells, but not to normal B cells or EBV-transformed B cell lines established from healthy donor PBMC (Fig. 2, data not shown). By contrast, mAbs 1A1-IgG and 1A7-IgG, despite binding to purified proteins in ELISA, did not show any binding to ROR1-positive cells or cell lines (data not shown), and further studies are needed to understand this-discrepancy. In agreement with our previous report using GaROR1 pAbs,9 the binding of 2A2-IgG revealed variations in the level of ROR1 expression (differences in MFI) among samples from individual CLL patients and, similarly, among samples from individual MCL patients (Fig. 2A and and2B).2B). However, ROR1 expression was uniform in all malignant B cells in each sample (Supplementary Figure 1S). Subsequent studies showed that the binding was specific to the malignant cells (CD5+CD19+ cells) from CLL and MCL patients; other PBMC subpopulations in CLL and MCL patients or CD19-positive cells from healthy donors were not recognized (Supplementary Figure 1S and data not shown). Titration of 2A2-IgG showed significant binding at a concentration as low as 1 ng/mL (~6.5 pM), suggesting a high affinity/avidity interaction (data not shown). In addition, mAb 2A2-IgG bound to all MCL cell lines tested (HBL-2, JeKo-1, Mino and Rec-1), but not to Burkitt lymphoma cell lines (Raji, Ramos and Daudi), EBV-transformed B cell lines (EBV-7475, RGC-EBV) or CLL cell lines (MEC-1, MEC-2 and 232-B4) (Fig. 2C and data not shown).
Similarly, like the parent mAb 2A2-IgG, immunotoxin BT-1 showed dose dependent binding to primary CLL and MCL cells and to the MCL cell lines JeKo-1 and Mino (Fig. 2A, ,2B2B and and2D).2D). The binding of BT-1 was restricted to the CD5+/19+ subpopulation in the PBMC samples from CLL and MCL patients (Fig. 2A and and2B).2B). No significant binding above background was seen with B cells, T cells or other PBMC from healthy donors, and ROR1-negative B cell lines RGC-EBV, Daudi, Raji and Ramos (Fig. 2D and data not shown).
To further investigate the affinity/avidity of 2A2-IgG and BT-1 in real time and label-free manner, we measured their interaction with hROR1 by surface plasmon resonance. First, hROR1-ECD or Fc-hROR1 (ligand) was immobilized on a CM5 sensor chip and mAb 2A2-IgG was run in the mobile phase (analyte). Regardless of the monovalent (hROR1-ECD) or bivalent (Fc-hROR1) nature of the ligand, 2A2-IgG showed a fast association (kon) followed by a slow dissociation (koff) resulting in observed equilibrium binding constants (KD) of 0.12 nM and 0.42 nM, respectively (Table 1, Supplementary Figure 3S). Next, in a reverse orientation, 2A2-IgG was captured as ligand on a CM5 chip, and hROR1-ECD or Fc-hROR1 were run as analytes. Whereas the association rates of the two analytes to the mAb did not differ much in this setting, monovalent hROR1-ECD dissociated faster than bivalent Fc-hROR1, presumably due to the lack of an avidity gain in the former. Consequently, monovalent binding showed a significant decrease in the observed KD values (0.79 nM vs. 32.6 nM) with approximately 40-fold lower affinity (Table 1, Supplementary Figure 3S).
ROR1-immunotoxin BT-1 revealed a fast association rate similar to that seen with 2A2-IgG in binding Fc-hROR1. However, in comparison to bivalent 2A2-IgG, monovalent BT-1 exhibited a more than 100-fold faster dissociation rate, resulting in an observed KD value of 64.9 nM (Table 1, Supplementary Figure 3S). Notably, the difference between the observed KD values of analytes 2A2-IgG and BT-1 was similar to that found for analytes Fc-hROR1 and hROR1-ECD, revealing the strong avidity gain conferred by bivalency when the antibody or the antigen was in solution phase.
Internalization of cell surface bound anti-ROR1 mAb measured as described before for pAbs.9 Both primary CLL cells and MCL cell lines exhibited partial internalization (approximately 30–45%) of cell surface bound 2A2-IgG (at 4°C) upon subsequent incubation at 37°C (Fig. 3A, ,3B,3B, and data not shown). Internalization was noticed as early as 30 min, reached a plateau by 2 h and no further increase was observed with prolonged incubation up to 6 h (data not shown). Internalization of cell surface bound mAbs was efficiently blocked in the presence of a chemical inhibitor, PAO, during incubation at 37°C (Fig. 3A, ,3B).3B). These results with mAbs are in agreement with our previous observation with GaROR1 pAbs using primary CLL cells.9 A similar level of internalization was observed when biotinylated 2A2-IgG was used, although the MFI was lower compared with non-biotinylated 2A2-IgG (data not shown). To further rule out the possibility that bound mAb dissociated from the cell surface during incubation at 37°C, we used fluorochrome (Atto-488) conjugated 2A2-IgG (Atto-488–2A2-IgG). This allowed simultaneous visualization of internalized (direct fluorescence) and surface retained (indirect fluorescence) mAb in non-permeabilized cells. Direct measurement of Atto-488 signals in JeKo-1 cells remained the same before and after incubation at 37°C. This is because both the internalized and cell surface retained antibody molecules are detected simultaneously, suggesting there was no dissociation of bound mAb. By contrast, indirect measurement using a fluorochrome-conjugated secondary antibody (Dy-GaM) that, in non-permeabilized cells, detects only Atto-488–2A2-IgG retained on the cell surface. As expected, there was a clear reduction in signal upon 37°C incubation (Fig. 3A), indicating that internalization took place. Additional experiments with another MCL cell line (Mino), as well as primary CLL and MCL cells showed similar levels of internalization of 2A2-IgG (Fig. 3B, data not shown).
Internalization of cell surface bound immunotoxin is a key requirement for mediating cytotoxicity. Results showed that cell surface bound BT-1 disappeared in MCL cell lines, as well as in primary CLL and MCL cells, following incubation at 37°C (Fig. 3C, data not shown). However, given the weaker binding kinetics of BT-1 (Table 1), along with the observations that disappearance was greater (60–80%) than that of 2A2-IgG and not prevented by PAO (Fig. 3C), it is likely that dissociation contributed to this disappearance. Nonetheless, a small amount of internalized BT-1 could potentially deliver toxin into target cells and induce apoptosis.
Three independent assays were performed to determine the ability of BT-1 to kill target cells. First, cell surface exposure of PS was detected by Annexin-V binding and was interpreted as a measure of apoptosis. ROR1-positive (JeKo-1) and ROR1-negative (Daudi and Raji) cell lines were incubated at 37°C in the absence or presence of different amounts of BT-1 (0.01 – 100 μg/mL) for 20, 46, 64 and 75 h. While BT-1 induced dose and time dependent killing (apoptosis) of JeKo-1 cells, Daudi and Raji cells were not killed (Supplementary Figure 4S). Similarly, another ROR1-positive cell line, HBL-2 exhibited 100% apoptosis when cultured with as little as 1 μg/mL (16 nM) BT-1 for three days, and the ROR1-negative cell line, RGC-EBV showed no increase from the background level of apoptosis when cultured with up to 100 μg/mL (1600 nM) of BT-1 for three days (Fig. 4). As expected, the untargeted small molecule drug camptothecin was able to induce apoptosis in all cell lines regardless of ROR1 expression, indicating that ROR1-negative cell lines are not refractory to apoptosis (data not shown).
Second, change in the mitochondrial membrane potential was measured as an indicator of apoptosis. HBL-2 and RGC-EBV cell lines were cultured for 24 h in the absence or presence of 1 μg/mL BT-1 and subsequently incubated with 12.5 nM DiOC2(3) and washed before flow cytometry. A drop in the red fluorescence intensity (as well as green fluorescence intensity, data not shown), which indicated a change in mitochondrial membrane potential (ΔmΨ), was noted in HBL-2 cells when cultured with BT-1, but not in RGC-EBV cells (Fig. 5A and and5B).5B). Lower concentrations of BT-1 (0.01 – 0.1 μg/mL) induced a lower drop of fluorescence intensity in HBL-2 cells (data not shown). Addition of 50 μM carbonyl cyanide 3-chlorophenylhydrazone (CCCP), a nonspecific mitochondrial membrane destabilizer, during incubation with DiOC2(3) served as a positive control and it induced change in mitochondrial membrane potential in both cell lines regardless of the presence of BT-1 (data not shown).
Third, intracellular activated caspase-3 was detected as an early indicator of apoptosis. HBL-2 and RGC-EBV cells were cultured 24 h in the absence or presence of 1 μg/mL BT-1, and subsequently stained with a rabbit mAb against human activated caspase-3. Almost all HBL-2 cells cultured with BT-1 exhibited the presence of activated caspase-3, whereas RGC-EBV cells showed only background levels of staining except for a small population of cells (Fig. 5C and and5D).5D). Collectively, these results demonstrated that BT-1 induced apoptosis in the ROR1-positive HBL-2 cell line.
To further confirm BT-1 as a targeted drug, four MCL cell lines expressing different levels of ROR1 (HBL-2, JeKo-1, Mino and Rec-1) were cultured in the absence or presence of different amounts of BT-1 (0.001 – 10 μg/mL) for three days. Four ROR1-negative cell lines (RGC-EBV, Daudi, Raji and 232-B4) served as controls. Based on Annexin-V binding, all ROR1-positive cell lines showed a dose-dependent increase in apoptosis and none of the ROR1-negative cell lines showed significant apoptosis over background (Fig. 6). The sensitivity to BT-1-induced apoptosis appeared to be influenced by target antigen density (ROR1 expression levels) on the MCL cell lines (Table 2). The HBL-2 cells had the highest ROR1 expression (based on both MFI and ABC values) and the EC50 of BT-1 was approximately 0.01 μg/mL (0.16 nM), followed by JeKo-1 at 0.1 μg/mL (1.6 nM) and Mino at 1.0 μg/mL (16 nM). The Rec-1 cells were highly sensitive to BT-1 (EC50 0.001 μg/mL (0.016 nM), but there was a discrepancy in the measurement of ROR1 expression. In repeated experiments, while the MFI values were low (Fig. 2C), the ABC values were high (Table 2) for Rec-1 cells and the reason for this inconsistency is not known. Next, as an independent correlate, the level of the anti-apoptotic protein, BCL2 was measured. The amount of intracellular BLC2 varied among the ROR1-positive MCL cell lines, and paralleled their sensitivity to BT-1 (Table 2). It is interesting to note that the highly sensitive Rec-1 cells have the lowest BCL2 level.
Targeted therapy is a novel approach in the treatment and control of cancers, and has the potential to enhance the therapeutic efficacy, minimize undesirable side effects and help achieve a better clinical outcome in patients. With respect to B cell malignancies, others and we have demonstrated uniform and restricted expression of ROR1 in CLL and MCL cells, and suggested its potential as a therapeutic target.9-11,36 The present study describes the development, characterization and validation of four mouse mAbs against hROR1. Subsequently, we focused on one of the two mAbs (2A2-IgG) that selectively bound ROR1-positive primary CLL and MCL cells as well as MCL cell lines and revealed the overall strongest reactivity with hROR1 and mROR1. Nonetheless, 2A2-IgG alone or after conversion to chimeric mouse/human IgG1 (data not shown) and three chimeric rabbit/human IgG1 against ROR130 failed to induce significant direct cytotoxicity, CDC or ADCC in primary CLL cells or ROR1-positive MCL cell lines. Such unarmed mAb-mediated effector functions have generally been demonstrated in cells with high target antigen density (e.g., CDC activity of rituximab on human B cells with > 40,000 CD20 molecules/cell).37 Our previous study estimated the ROR1 density on CLL cells to be between 2,500 and 7,000 molecules/cell,9 which may be below the threshold needed for unarmed mAbs to mediate effector functions.38,39 It is also possible that the effector functions of unarmed mAbs against ROR1 are influenced by epitope location.40
Knowing that cell surface ROR1 mediated internalization of bound antibody in primary CLL cells and MCL cell lines, the use of a ROR1-immunotoxin was sought to potentially overcome this limitation. The induction of profound apoptosis by mAb 2A2-derived ROR1-immunotoxin BT-1 in dsFv-PE38 format in ROR1-positive MCL cell lines, but not in ROR1-negative B cell lines validated that ROR1 can function as a therapeutic target. The specific cytotoxicity induced by BT-1 was confirmed by three independent measurements: PS exposure to cell surface, change in mitochondrial membrane potential and activation of caspase-3, events that have been well documented in apoptotic cell death. The maximum effect was seen after approximately 3 d and this is in agreement with other studies using PE38-based immunotoxins.16 The dose-response curves depicted differences in BT-1 sensitivity among different MCL cell lines, and this could reflect cell specific intrinsic factors as regulators of apoptosis.
A number of parameters determine the efficacy of immunotoxins. First, the affinity of a mAb or a mAb fragment (e.g., dsFv) to its target antigen largely determines the stability of the antibody/antigen complex on the cell surface. A head-to-head comparison of the observed KD values of bivalent 2A2-IgG (0.42 nM) and 2A2-derived monovalent BT-1 (64.9 nM) binding to immobilized Fc-ROR1 revealed an approximately 150-fold difference in BT-1 binding. Assuming that similar binding might occur with cell surface ROR1 and given that the half-life of the antigen/antibody interaction equals ln2/koff, the estimated cell surface retention of BT-1 would be 19 sec compared with 34 min for 2A2-IgG. Although the dose-response curves with BT-1 showed robust killing of ROR1-positive MCL lines at low concentration (with EC50 values of 16 pM to 16 nM), there is likely room for improvement. Other PE38-based immunotoxins that target different antigens on malignant B cells (CD19 and CD22) or mesothelioma (mesothelin) showed higher potency presumably due to higher affinity, i.e., lower dissociation rate constants, for the respective antigens, resulting in longer cell surface retention.41 Currently, we are looking into different antibody engineering strategies to improve the affinity or avidity of BT-1. Alternatively, a new panel of chimeric rabbit/human mAbs to ROR1 that was selected by phage display in the monovalent Fab format for high affinity binding to ROR1 and reveals dissociation rate constants up to 120-fold lower than BT-1,30 could be employed for the next generation of ROR1-immunotoxins.
Second, the target antigen density on the cell surface could influence immunotoxin efficacy. Our data showed an approximately 10-fold difference in EC50 values among ROR1-positive MCL cell lines (0.016 nM for Rec-1, 0.16 nM for HBL-2, 16 nM for JeKo-1 and 16 nM for Mino) and overall this paralleled ROR1 expression levels among these cell lines. A correlation between target antigen density and apoptosis induction was also reported for CD22- and FCRL1-immunotoxins.16,42 However, preliminary results indicate that Rec-1 cells expressed the lowest level of the anti-apoptotic protein BCL2 compared with the other MCL cell lines we tested. It is conceivable that BCL2 levels in addition to the level of other anti-apoptotic proteins (e.g., MCL1, survivin, X1AP) can influence the efficacy of BT-1 as has been reported for other immunotoxins.43,44
Third, our epitope mapping studies indicated that 2A2-IgG (the parent antibody of BT-1) bound to the membrane distal Ig-domain of ROR1. Immunotoxins generated from antibodies recognizing different epitopes on the same antigen showed significant differences in their potency in vitro.45,46 An immunotoxin binding to a membrane proximal epitope on the target antigen, CD2, induced higher cytotoxicity than those binding to distal epitopes.47 Therefore, it may be worthwhile to generate ROR1-immunotoxins based on mAbs recognizing different epitopes and compare their abilities to kill ROR1-expressing cells.
Fourth, the rate of internalization of surface bound immunotoxin could affect its efficacy. A recent study compared the efficacy of two immunotoxins, RFB4 (Fv)-PE38 also known as BL22 (anti-CD22) and FMC63 (Fv)-PE38 (anti-CD19) on lymphoma cell lines. Despite lower cell surface expression of CD22, rapid internalization of large amounts of BL22 made CD22 a better target than CD19 present in high density.48 There was clear evidence for BT-1-mediated apoptosis in ROR1-positive cell lines within 24 h, but how much BT-1 was actually internalized is unknown. Further studies on the membrane dynamics of ROR1, intracellular trafficking of BT-1 and its degradation rate are required to better understand and improve the efficacy of next generation ROR1-immunotoxins.
Despite the profound cytotoxicity induced by BT-1 in vitro, the surface plasmon resonance data indicated relatively fast dissociation of the monovalent BT-1 compared with its parent antibody, and it is likely that this could be a limitation in exploring its in vivo activity. We are currently working on generating different formats of ROR1 immunotoxins with a goal to enhance the affinity/avidity, and minimize potential immunogenicity. The efficacy of improved immunotoxins can be tested in vivo using MCL cell lines as xenografts in nude mice. The cytotoxicity studies described here were based on MCL cell lines and, unlike primary CLL and MCL cells, they do not undergo spontaneous apoptosis in vitro. A more clinically relevant system for evaluating the ability of ROR1-immunotoxins to kill primary CLL and MCL cells may be a recently developed mouse model with adoptively transferred human primary CLL cells that retain ROR1 expression in vivo.49
In summary, as a proof of concept, we have demonstrated that a ROR1-immunotoxin can induce apoptosis resulting in robust killing of ROR1-expressing MCL cell lines in vitro. Owing to its exquisite specificity, BT-1 could function as a leukemia/lymphoma specific therapeutic agent in vivo, and unlike unarmed mAbs, would not depend on the effector machinery such as complement proteins, NK cells, macrophages or T cells as these may not be optimally available in cancer patients. BT-1 provides a platform for next generation ROR1-immunotoxins with defined modifications in both antibody and toxin portion designed to enhance affinity and efficacy, as well as minimize immunogenicity.50,51 In addition, combination strategies with small molecule inhibitors of anti-apoptotic proteins could help to mitigate the opposing effects and boost the efficacy of BT-1 and next generation ROR1-immunotoxins.
The authors declare no competing financial interest.
This work was [in part] funded by the Intramural Research Program of the National Institutes of Health [National Cancer Institute and National Heart, Lung and Blood Institute]. We thank Mr. Michael G.Kennedy and Dr. Jiahui Yang for the generation of recombinant ROR1 proteins, Dr. Anders Rosen for providing the 232-B4 cell line and Dr. William G. Telford and Ms. Veena Kapoor for maintaining the flow cytometry core facility.
Previously published online: www.landesbioscience.com/journals/mabs/article/19870