The Development of a Recombinant scFv Monoclonal Antibody Targeting Canine CD20 for Use in Comparative Medicine

Abstract

Monoclonal antibodies are leading agents for therapeutic treatment of human diseases, but are limited in use by the paucity of clinically relevant models for validation. Sporadic canine tumours mimic the features of some human equivalents. Developing canine immunotherapeutics can be an approach for modeling human disease responses. Rituximab is a pioneering agent used to treat human hematological malignancies. Biologic mimics that target canine CD20 are just being developed by the biotechnology industry. Towards a comparative canine-human model system, we have developed a novel anti-CD20 monoclonal antibody (NCD1.2) that binds both human and canine CD20. NCD1.2 has a sub-nanomolar Kd as defined by an octet red binding assay. Using FACS, NCD1.2 binds to clinically derived canine cells including B-cells in peripheral blood and in different histotypes of B-cell lymphoma. Immunohistochemical staining of canine tissues indicates that the NCD1.2 binds to membrane localized cells in Diffuse Large B-cell lymphoma, Marginal Zone Lymphoma, and other canine B-cell lymphomas. We cloned the heavy and light chains of NCD1.2 from hybridomas to determine whether active scaffolds can be acquired as future biologics tools. The VH and VL genes from the hybridomas were cloned using degenerate primers and packaged as single chains (scFv) into a phage-display library. Surprisingly, we identified two scFv (scFv-3 and scFv-7) isolated from the hybridoma with bioactivity towards CD20. The two scFv had identical VH genes but different VL genes and identical CDR3s, indicating that at least two light chain mRNAs are encoded by NCD1.2 hybridoma cells. Both scFv-3 and scFv-7 were cloned into mammalian vectors for secretion in CHO cells and the antibodies were bioactive towards recombinant CD20 protein or peptide. The scFv-3 and scFv-7 were cloned into an ADEPT-CPG2 bioconjugate vector where bioactivity was retained when expressed in bacterial systems. These data identify a recombinant anti-CD20 scFv that might form a useful tool for evaluation in bioconjugate-directed anti-CD20 immunotherapies in comparative medicine.

Fig 1. Isolation of a mouse CD20-specific monoclonal antibody that can recognize human and canine CD20 protein.

(A). Amino acid sequence of human and canine CD20 surrounding the Rituximab epitope. The asterisks highlight divergence between human (hCD20) and canine (cCD20) proteins. (B). Immunoblotting of recombinant canine CD20 protein. Bacterially expressed his-tagged canine CD20 protein (amino acids 140–190) was purified by nickel chromatography and after electrophoresis without and with DTT, immunoblotted with MAB NCD1.2. The laddering observed without DTT presumably reflects inter-molecular di-sulfide bonds through the two cysteine residues. (C and D). (C). Immunoblotting of authentic canine CD20 protein. 3132 canine lymphoma cells were lysed with different lysis buffers (as indicated) to determine optimal extraction buffer to immunoblot endogenous CD20 protein. (D). 3132 canine lymphoma cells were subjected to chemical fraction (Fr1-Fr4, as indicated) to isolate compartments to determine the dominant localization of CD20 protein in 3132 cells. (E and F). Immunoblotting of human CD20 protein. Transfection of human GFP-CD20 into H1299 cells demonstrated that NCD1.2 can bind to human CD20 (Left panel) with and anti-GFP monoclonal antibody as a control for protein expression and relative molecular mass (right panel).

Fig 2. Definition of the relative affinity of the NCD1.2 MAB towards the epitope peptide.

(A and B). NCD1.2 was titrated into reactions containing canine CD20 peptide on the solid phase in the absence or presence of DTT to evaluate potential oxidation effects on epitope binding. An Octet RED biolayer interferometry system that measures binding to the sensor tip as a wavelength shift (in nm) in real time. The assay data were processed using Data Analysis (version 6.3—Forte Bio) to obtain kinetic values as in the materials and methods and tabulated as K_d, K_on and K_diss.

Fig 3. Comparison of NCD1.2 and Rituximab.

(A). NCD1.2 and Rituximab were titrated into reactions containing fixed amounts of fluorescent human CD20 peptide (FITC; 30 nm peptide) and relative binding affinities were compared to each other in order to determine how the epitopes for the two monoclonal antibodies might differ. The data are plotted as changes in polarization as a function of increasing antibody concentration. (B). The human SUDHL4 B-cell lymphoma cell line expressing CD20 was incubated with Rituximab conjugated by DyLight 488 (10 μg/ml). The non-labeled Rituximab was added in increasing concentrations to samples 2–5. The non-labeled monoclonal antibody NCD1.2 was added to samples 6–8. The intensity of Rituximab-DyLight 488 was measured using flow cytometry.

Fig 4. Expression of CD20 protein in clinically-derived canine cell populations.

The highlighted cell sample populations were isolated and processed as indicated in the materials and methods. Expression of CD20 (and CD21) in the population of cells were analyzed by FACS and include (A) peripheral blood; (B) peripheral T-cell lymphoma; (C) Medium sized B-cell lymphoma; and (D) Diffuse Large B-cell lymphoma.

Fig 5. Expression of CD20 protein in formalin fixed paraffin imbedded canine cancer tissue.

The indicated formalin-fixed, paraffin-embedded tissues were processed using immunohistochemistry as indicated in the materials and methods; and include representative images: (A) NCD1.2 staining in diffuse large B cell lymphoma (20x); (B) polyclonal anti-cd20 rabbit antibody staining in DLBCL (20x); (C) NCD1.2 staining in marginal zone lymphoma (20x); and (D) lack of NCD1.2 staining in peripheral T-cell lymphoma (10x); with infiltrating normal B-cells that are CD20+ highlighted by the arrow. The staining in brown highlights the position of the NCD1.2 reactive protein with nuclei stained in blue.

Fig 6. Strategy for cloning the variable light and heavy chains from the NCD1.2 hybridoma cell to create a NCD1.2 scFv-phage display library.

RNA is extracted from hybridoma cell lines, followed by cDNA production. The cDNA was used as a template for the PCR-amplification of heavy and light chain variable regions using a diverse template primer set in order to capture the expressed sequences. The purified light and heavy chains were spliced together using PCR to create a 800 bp overlap product. This product was digested with SfiI, ligated into pCOMB3xSS, and transformed into TG1 cells. These cells are then be used to propagate the scFv-phage library for isolation of bioactive scFv-gIII fusion proteins.

Fig 7. Amplification of heavy and light chains using PCR primer sets.

RNA was isolated from NCD1.2 hybridoma cells and PCR amplified products (using the numbered oligonucleotide primers sets below each fig (decoded in the materials and methods) from cDNA were isolated on a 1.5% agarose gel: (A). Variable light chain amplification with 5’ sense and 3’ reverse primer sets: Lane L, ladder; Lane 1, negative control; Lanes 2–9, incorporate 5’ primers MSCVK1-MSCVK8; lanes 10–18 incorporate 5’ primers MSCVK9-MSCVK17. The 3’ primer reverse sets in each reaction were MSCJK12-BL, MSCJK4-BL, and MSCJK5-BL. (B). Variable heavy chain amplification with 5’ sense and 3’ reverse primer sets: Lane L, ladder; Lane 1, negative control; Lanes 2–11, incorporate 5’ primers MSCVH1-MSCVH10 and lanes 12–20 incorporate 5’ primers MSCVH11-MSCVH19. The 3’ primer reverse sets in each reaction were MSCM-B MSCG3-B MSCG1ab-B. The arrow highlights the position of the variable heavy or light gene. The amino acids sequences below each gel (from primers 4, 5, 13, and 15 for the light chain and primers 8, 9–13, 15, and 16 for the heavy chain) highlight the possible sequences of the framework regions to be expected in the sequences of the isolated, bioactive scFv.

Fig 8. Isolation of bioactive recombinant scFv targeting CD20.

(A). Soluble scFv fragment binding to its antigen. Biotinylated CD20 peptide (0.1 μg/ml in 50 μl) was coated onto streptavidin (1 μg/ml in 50 μl) coated solid phase and used as an antigen in the selection of scFv phage that bound to the CD20 epitope as indicated in the materials and methods. As the selection in rounds progressed, there was a reduction in the amount of antigen in the selection and an increased stringency of washing. (B). Colonies (96) from rounds 3 and 4 were grown in 200 μl of LB media and induced overnight at 30°C to overproduce soluble scFv fragments. Following a freeze-thaw and sonication protocol to release scFv from the bacterial debris, the clarified lysate was assayed in the ELISA with peroxidase conjugated to protein A. The binding was quantified using TMB and was measured in optical density at 450nm. Five high affinity scFv were identified (highlighted 1–5). However, their instability precluded their routine use (data not shown). (C). Phage producing plaques from rounds 3 and 4 were grown overnight, the scFv-phage in the supernatant was PEG precipitated, and assayed in an ELISA using biotinylated CD20 peptide. Active scFv-gIII-fusion phage was detected using an M13-phage antibody and binding was quantified using TMB and plotted as optical density (450 nm). Eight representative scFv-gIII fusion phage are highlighted where two bioactive recombinant phage (3 and 7) were isolated.

Fig 9. Amino acid sequences of scFv-3 and scFv-7.

(A and B). The amino acid sequences of the light chain and heavy chain framework and CDR regions are as indicated. The CDRs of the light chain are highlighted in red, since the sequences diverge in CDR1 and CDR2 of the two light chains, but are identical in the CDR3 of the light chain. (C). The full amino acid sequence of scFv-3 and scFv-7 including the linker is highlighted. Based on the amino acid sequences, the scFv-3 and scFv-7 light chains could have been amplified using any of the primers 4,5,13, or 15 (Fig 7A (in orange)). The heavy chain amino acid sequences also represent different PCR amplicons but with identical CDRs; the scFv-3 heavy chain could have been amplified using primers 8, 9, or 15, highlighted in green (Fig 7B). The scFv-7 heavy chain could have been amplified using primers 8, 9, or 10, highlighted in green (Fig 7B). (D). RT-PCR was applied using primers directed to the divergent scFv-3 and scFv-7 light chains (from the divergent CDR1 sequences (in red) to Framework 4) expressing from the NCD1.2 hybridoma cell line to establish that the light chain mRNA for scFv-3 and scFv-7 are both expressed to similar levels, as defined by the amount of PCR products after 15, 20, or 30 PCR cycles. The size of the CDR1-FR4 PCR product is approximately 300 bp and the 200 bp ladder is Hyperladder (Bioline). These data indicate that the hybridoma produces two light chains with an identical CDR3.

Fig 10. Predicted hypermutation events in scFV3 and 7.

(A-C) The predicted hypermutation events that derived the scFv-3/7 (from the IMGT/V-QUEST web site).

Fig 11. Bioactivity of recombinant scFv-3 and scFv-7 secreted from CHO cells and produced in bacteria as a CPG2-bioconjugate.

(A) The indicated scFv (3 or 7) was cloned into pCDNA3.1 containing a leader sequence (amino acids MGGS) for targeted secretion into the media of tissue cultured CHO cells. (i) Binding was measured on left panel after dilution of the supernatants (1:40 or 1:80) to optimize binding against His-tagged CD20 protein; (ii) right panel the optimized supernatants of scFv-3 and scFv-7 were assayed against biotinylated CD20 peptide as indicated in the materials and methods. Antibody scFv binding to antigen was detected using peroxidase conjugated protein A and resolved with TMB-based ELISA at an OD of 450nm. Controls included media only or PBS. (B). A schematic of the scFv-3 and scFv-7 CPG2 fusion protein. (C). Optimized purification of scFv-3/7 from bacteria. Bacteria were grown and scFv-3 or scFv-7 CPG2 bioconjugates were induced with IPTG. Following lysis with 100 mM or 1 M NaCl lysis buffer, the samples were separated into soluble and insoluble fractions. These fractions were mixed with SDS loading buffer, were separated on an SDS gel, and stained with Coomassie blue. The Left panel (i) displays the solubility (S) or insolubility (I) of scFv-3 and scFv-7 from bacterial expression systems in lysis buffer containing the indicated NaCl concentrations. The arrow highlights the position of the scFv-CPG2 fusion protein. The NI lane represents the amount of soluble scFv-3 recovered from the S fraction after nickel chromatography, which was negligible. The right panel (ii) shows the relative purify of the scFv-3 and scFv-7 (and scFv-8 as a control antibody) after lysis using stabilizing lysis buffer (as in Methods) and followed by nickel affinity chromatography. The relative expression and purity of CPG2 alone is shown by comparison to highlight its enhanced yield relative to the scFv-CPG2 fusion. The insolubility remains a problem from bacterial expression systems, as the total synthesis of scFv-CPG2 and CPG2 alone is relatively similar using whole cell lysis buffer (right panel (iii). (D). Bioactivity of affinity purified scFv-3 and scFv-7 from bacteria. The normalized affinity-purified scFv fractions were assayed for binding to the indicated canine CD20 peptides (peptide 32 and peptide 38), where the core epitope resides (in gold shade). scFv-3:CPG2 was more active than scFv-7:CPG2 on both peptide 32 and peptide 38. The enhanced binding of scFv-3:CPG2 as a bioconjugate is consistent with its enhanced binding when secreted as a single chain from CHO cells and when fused to M13 gIII protein.