Anti-CEA Delta CH2 Antibody Design and Production

The humanized anti-CEA hT84.66-M5A (M5A) mAb was engineered into a CH2 domain-deleted (∆CH2) mAb fragment format based on the anti-TAG-72 ∆CH2 antibody [23, 33]. The IgG1 heavy chain CH2 domain was replaced with a linker consisting of serine (S) and glycine (G) amino acids to join the IgG1 lower hinge to the CH3 domain. Two different linkers named Flex (APGGGSSGGGSG), and TAG (GGGSSGGGSG) were incorporated as previously described [10, 13]. Modifying the IgG hinge design of Glaser et al., four hinge and linker variants were developed having two (C2), three (C3) and five (C5) cysteine disulfide bridges (M5A∆CH2-C2_Tag, M5A∆CH2-C2_Flex, M5A∆CH2-C3_TAG, and M5A∆CH2-C5_Flex, respectively) as shown in Fig. 1 [10].

cDNA encoding the gene constructs were synthesized by GeneArt (ThermoFisher Scientific, MA). The individual pairs of light and heavy chain genes were subcloned into the pEE12/6 dual vector GS expression system (Lonza Biologics, Switzerland). Plasmids encoding the individual delta CH2 antibody constructs were transiently expressed using the ExpiFectamine™ 293 Transfection Kit (Gibco) as per manufacturer instructions. The cells were grown in Expi293 expression media at 125 rpm, 37 ˚C and 8% CO2 in a humidified incubator and harvested on day 6 post-transfection.

Purification of M5A∆CH2 Antibodies

The cell harvests were clarified by centrifugation (3000 g for 10 min) and filtered to remove particulates. The cell-free harvest was treated with the strong anion AG1-X8 resin (Bio-Rad, Hercules, CA) (5% w/v) by incubation on a roller assembly at 4˚C overnight and sterile filtered to provide a clarified feed stream. Chromatography was performed using an NGC chromatography system (BioRad) controlled and recorded using ChromLab software (Bio-Rad). For affinity purification an AP-1 column (Waters, MA) was packed with 7 ml of Capture Select™ FcXP affinity matrix (ThermoFisher Scientific). The column was pre-equilibrated with 20 column volumes (CV) of phosphate buffered saline (PBS) pH 7.4 at 2 ml/min. The individual harvests were loaded onto the column at a flow rate of 1 ml/min. The column was washed with 5 CV PBS, 10 CV of a high salt buffer (0.02 M sodium phosphate, 0.02 M sodium citrate, 0.5 M NaCl, pH 7.5) and 5 CV of PBS. The M5A∆CH2 constructs were eluted with 0.1 M glycine, pH 3.0 buffer. Eluted protein was neutralized with 0.5 M [2-(N-morpholino)ethanesulfonic acid] (MES), pH 8.0 buffer to pH 6.5. The column was cleaned and sanitized with 1% phosphoric acid and 2 M guanidine hydrochloride and stored at 4˚C in 20% ethanol.

Ceramic Hydroxyapatite Type 1, 20 µm (CHT) (Bio-Rad) chromatography was used as polishing step to remove aggregates at a flow rate of 5 ml/min (except where stated otherwise). A 7 ml CHT column was pre-equilibrated with 10 CV of 0.05 M MES, 0.01 M potassium phosphate, pH 6.5 buffer. The purified antibody was loaded at 2.5 ml/min followed by 2 CV of wash buffer (0.05 M MES, 0.05 M potassium phosphate, pH 6.5). A 30 CV linear gradient of 100% wash buffer to 100% elution buffer (0.05 M MES, 0.2 M potassium phosphate, pH 6.5) was used to elute the protein. Monitoring elution by A280, peaks were collected and analyzed for the presence of aggregates by high performance liquid chromatography-size exclusion chromatography (HPLC-SEC) using a Superdex 200 10/300 column (Cytiva, Wilmington, DE). The peaks containing the monomeric form were combined, buffer exchanged with PBS using 10 kDa MWCO protein concentrators (ThermoFisher Scientific) and sterile filtered. M5A∆CH2 antibody fragments were stored at a concentration range of 6.5–7.5 mg/ml in PBS at 4˚C.

Biochemical Analysis

Antibody samples were analyzed by SDS-PAGE under non-reducing and reducing conditions on 10% Mini Precast Protein Gels (Bio-Rad). Three micrograms of antibody samples were mixed with loading buffer with or without DTT reducing agent. The sample were heated for 5 min at 95˚C before loading and electrophoresed at 200 V × 30 min. The gels were imaged on a ChemDoc imaging system (Bio-Rad) and analyzed using the Image-Lab software (Bio-Rad). To compare the thiol stability of M5A∆CH2 constructs, the samples were incubated with TCEP for one hr at room temperature. TCEP: protein molar ratios of 30:1, 15:1, 7.5:1, and 3.5:1 were used. The samples were run on SDS-PAGE under non-reducing conditions. Stability was assessed at serial time points by HPLC-SEC analysis. Anti-CEA immunoreactivity was confirmed by incubating 10 µg of antibody with 50 µg of soluble CEA, (37˚C for 30 min) and analyzing for the formation of a 300 kilodaltons (kDa) antibody-antigen complex by HPLC-SEC [17].

Surface plasmon resonance (SPR) assays were performed on Biacore X100 (Cytiva) by using recombinant human CEA biotinylated (RayBiotech, Corners, GA) immobilized on sensor chip SA (Cytiva) at a concentration of 5 µg/ml. M5A and M5A∆CH2 mAb constructs were titrated at 8 concentrations (1000, 500, 250, 125, 62.5, 31.25, 15.62, and 7.8 nM). Each run had 300 s contact time with the analyte, 900 s dissociation time with a flow rate of 30 µl/min followed by 2 regeneration steps of 6 M guanidine hydrochloride with a contact time of 60 s each. The sensograms were analyzed for calculating equilibrium dissociation constant (KD) using Biacore evaluation software incorporating the 1:1 binding model.

Conjugation, Radiolabeling and Immunoreactivity

The M5A∆CH2-C5 antibody was conjugated with 1,4,7,10 tetraazacyclododecane-1,4,7-triacetic acid trisodium salt-vinyl sulfone (DO3A-VS) as previously described [17]. Briefly, 2 mg of M5A∆CH2-C5 was added to 376 µl of PBS and 24 µl of 10 mM TCEP in a microcentrifuge tube under argon, and incubated at room temperature with rocking for 2 h. The reduced antibody was reacted with 13 µl of DO3A-VS (10 mg/ml stock) and incubated at room temperature with rocking for 2 h. Unconjugated DO3A-VS and TCEP were removed by diafiltration with 0.25 M ammonium acetate, pH 7 (25 DV) using 10 kDa MWCO ultrafiltration membrane in an Amicon stirred cell (Millipore, MA). The DO3A-VS-M5A∆CH2-C5 was radiolabeled with [64CuCl] (3D Imaging, Little Rock, AR, specific activity 14.1 µCi/µg, in 1 M HEPES for 1 h at 43˚C) The radiolabeling efficiency was 98% by instant thin-layer chromatography. The [64Cu]Cu-DO3A-VS-M5A∆CH2-C5 was purified by HPLC-SEC. Incubation with soluble CEA (20 molar excess) showed > 95% immunoreactivity by an in vitro molecular weight shift assay and stability study showed the product was stable at least to 72 h by HPLC-SEC as shown in Fig. 2d [17].

Fig. 2

Biochemical and stability analysis of the M5AΔCH2 antibody constructs. a The purification of the M5AΔCH2 antibody constructs resulted in a single peak of the expected molecular size by HPLC-SEC analysis and (b) were stable up to 9 months. c Analysis of purified M5AΔCH2 antibodies on SDS-PAGE under reducing (left panel) and non-reducing conditions (right panel). The M5AΔCH2 series showed varying expression of two different isoform forms on the non-reduced gel (Form A and Form B). d The M5AΔCH2-C5 was mildly reduced, conjugated with DO3A-VS metal chelator, labeled with [64Cu] and shown to be immunoreactivity to soluble human CEA by a shift in molecular mass on HPLC-SEC.

Animal Model and Study Design

All applicable institutional and/or national guidelines for the care and use of animals were followed. All mice were handled in the City of Hope (COH) animal care facility in compliance with COH Institutional Animal Care and Use Committee guidelines, in accordance with the National Institute of Health Office of Laboratory Animal Welfare guidelines. Two animal models were employed bearing subcutaneous colorectal cancer tumors. Six 6-week old female athymic mice were injected with human colorectal cancer LS174T tumors (106 cells in 100 µl per mouse) and six 6-week old female immunocompetent transgenic-CEA (Tg-CEA) mice were injected with MC-38 tumors transfected with the human CEA gene in the flank as previously described [6, 18]. After 12 days, four mice from each group were selected based on the tumor size and injected via the tail vein with 100 µCi/10 µg of [64Cu]Cu-D03A-VS-M5A∆CH2-C5.

Imaging and Biodistribution

Serial imaging studies were conducted using β-cube and X-cube (MoleCubes, Ghent, Belgium) for PET and computed tomography (CT) scans, respectively. Mice were kept sedate under isoflurane anesthesia during each imaging session. In both animal model groups, two of the four mice were designated for serial immunoPET imaging at 0, 4, 24, and 48 h post-injection. PET and CT scan images were co-registered using manufacturer-provided software.

Blood clearance was measured by microcapillary sampling of 5 µl of blood from the tail vein at 0, 2, 4, 24, and 48 h post-injection and radioactivity was counted using a calibrated PerkinElmer gamma counter. After the last blood sample or image was acquired at 48 h, all animals were euthanized, necropsy performed, organs weighed (tumor, blood, heart, lung, liver, stomach, small and large intestine, spleen, kidneys, right quadricep muscle, and carcass) and counted for radioactivity. All data are reported as mean values and have been corrected for radioactive decay back to the time of injection, allowing organ uptake to be reported as percent of the injected dose per gram (% IDg–1) with standard errors. All statistical analyses were conducted using Prism version 9 (GraphPad Software, San Diego, CA). A two-phase decay nonlinear curve fit with a constrain (plateau = 0) was used to calculate the half-life.