Tag-free, specific conjugation of glycosylated IgG1 antibodies using microbial transglutaminase

We present an efficient approach for tag-free, site-specific conjugation of a fully glycosylated antibody using microbial transglutaminase (mTG). We created variants of trastuzumab where a single surface-exposed residue of the human crystallizable fragment had been substituted to glutamine, with the objective of enabling site-specific mTG-mediated conjugation with primary amine payloads. MTG reactivity was determined by conjugation to an amino fluorophore, demonstrating effective tag-free conjugation at the newly introduced I253Q site. The conjugation of one payload per antibody heavy chain was confirmed by mass spectrometry. We further demonstrated two-step mTG/click chemistry-based conjugation of I253Q trastuzumab with monomethyl auristatin E. Cytotoxicity and specificity of the resulting antibody–drug conjugate were indistinguishable from trastuzumab conjugated by another method although binding to the neonatal Fc receptor was impaired. The resulting fully glycosylated ADC is unique in that it results from minimal modification of the antibody sequence and offers potential for application to cellular imaging, fluorescence microscopy, western blotting or ELISA.

mM) and 200 µL polyethyleneimine (PEI) solution. DNA-PEI complexes were mixed by vortexing and incubated for at least 15 min at room temperature. For transient transfection, freshly cultured cells were adjusted to a cell titer of 2.5 × 10 6 cells for a final volume of 30 mL. The DNA-PEI solution was added dropwise to the cells while gently shaking. The cells were supplemented with 825 µL of 20% (w/v) tryptone 16-18 h post-transfection. Antibody expression was performed for 5 days at 37 °C, 110 rpm, and 8% CO 2 .
Production was stopped by centrifugation at 1,000 × g for 10 min and the supernatant was sterilized by filtration before purification. The cell culture supernatant was diluted with an equivalent volume of PBS and applied at a flow rate of 1 mL/min to a 1-mL Hi-Trap Protein A column (GE Healthcare) pre-equilibrated with at least 5 CV of PBS (pH 7.4). The column was subsequently washed with 5-10 column volumes of PBS buffer and the protein was recovered by isocratic elution (0.1 M sodium citrate pH 3) in 1-mL fractions were recovered into tubes containing 200 µL of 1 M Tris-HCl (pH 9). Protein fractions were pooled according to the A 280 peak and dialyzed against 5 L of PBS buffer (pH 7.4) overnight at 4°C with stirring. Protein was concentrated to >2 mg/mL using an Amicon 50 kDa MWCO (Merck Millipore). Protein concentration was determined at A 280 using a molar extinction coefficient of 2.1 x10 5 M -1 cm -1 and an approximate molecular weight of 150 kDa. Protein fractions were resolved by migration on 15% SDS-PAGE analysis to assess protein purity.

Production and purification of mTG
Microbial transglutaminase was produced and purified as described previously 1 . The plasmid pET20b-FRAP-mTG was kindly provided by Professor M. Pietzsch (Martin-Luther-Universität, Halle-Wittenberg, Germany) 2 . The specific activity was determined by the hydroxamate test and purity was assessed by SDS-PAGE analysis 3 . Specific activity was between 5-30 U/mg and purity was over 80% (Fig. S4).

Click-chemistry reactions
BCN-conjugated trastuzumab was purified over protein A spin column (GE Healthcare) according to the manufacturer's instructions. Azide-PEG 3 -vc-PAB-MMAE (MedChemExpress, 5 eq. per site) was added to the purified BCN-conjugated trastuzumab (1 mg/mL) and incubated for 12 h at room temperature. Excess unconjugated MMAE was removed using an Amicon Ultra-0.5 (MWCO 10 kDa, Millipore Sigma), washing three times with PBS.

Visualization of reactivity by SDS-PAGE analysis
SDS-PAGE analysis was performed for rapid visualization of reactivity with dansylcadaverine. An aliquot (5 µg) of reactions and their corresponding negative control in reducing SDS gel loading dye were denatured at 98 °C for 10 min and resolved by migration on 10% SDS-PAGE. Gels were run at 200 V until elution of bromophenol blue. Before Coomassie staining, fluorescence was revealed using a GelDoc-It imaging system (UVP Imaging) with an emission filter at 590 nm.

Hydrophobic interaction chromatography (HIC)
HIC was performed to assess the degree of conjugation (DoC) of the antibody variants with dansylcadaverine and BCN. A 2.5 mm × 4.6 mm × 35 mm TSKgel Butyl-NPR column (Tosoh Bioscience), fitted on an HPLC 1260 Infinity device (Agilent) equipped with a DAD detector was used for the analysis. Samples (35 µg) were injected onto the column and separation was performed using a flow rate of 0.9 mL/min using a linear gradient of 0% to 100% eluent B over 35 min (eluent A: 1.5 M ammonium sulfate, Tris-HCl pH 7.5, eluent B: Tris-HCl pH 7.5) to determine reactivity of all variants with dansylcadaverine. BCN and MMAE conjugation to I253Q and WT trastuzumab were analyzed using a gradient of 0 to 100% over 20 min. Detection was performed at 220 nm and the area under the peak was used to determine the degree of conjugation.

HER2 receptor cell binding assay
Flow cytometry was used to determine the binding affinity of glutamine-reactive variants of trastuzumab on HER2-overexpressing SK-BR-3 cells. Dilutions, incubation and washing steps were performed at 4 °C with PBS (pH 7.4) containing 1% BSA. Serial dilutions of the antibody (250 nM to 0.12 nM) were incubated for 1 h with 2 × 10 5 cells. Phycoerythrin-labeled goat anti-human IgG Fc antibody (1:75 dilution) (Invitrogen) was added and incubated for 30 min. A BDInflux cell sorter and BD FACS software were used to determine cell fluorescence over 5 × 10 4 events.

In vitro cytotoxicity
SK-BR-3 and HeLa cells were used to determine the cytotoxicity and specificity of the I253Q trastuzumab glutamine variant conjugated to MMAE. In a 96-well plate, cells were seeded at 5 000 viable cells/well the day before the experiment. Serial dilutions (20 nM to 0.02 nM) of antibodies were incubated with either of the cell types for 72 h at 37 °C, 5% CO 2 . The CellTiter96 ® AQ ueous cell proliferation assay (Promega) was used to assess cell viability according to the manufacturer's instructions.

FcRn binding
Affinity determination was performed on the Octet® RED96 system (FortéBio, Molecular Devices) using Octet® anti-human Fab-CH1 2 nd generation (FAB2G) biosensors. Sensors were soaked in PBS pH 7.4 for 10 min, and subsequently loaded with 10 µg/ml of the antibody of interest in PBS pH 7.4 until a response of 0.7 to 1 nm was reached, followed by 2 min of quenching in kinetics buffer pH 6.0 (FortéBio). Association of decreasing concentrations of FcRn (kindly provided by Merck KGaA (Darmstadt, Germany)), ranging from 64 nM -4 nM in a two-fold dilution series, was measured in PBS pH 6.0 for 7 min. As a negative control, kinetics buffer pH 6.0 was used instead of FcRn solution. Due to the pH-dependent binding mechanism of FcRn, dissociation was measured in PBS pH 7.4 for 10 min. All measurements were performed at 30 °C and 1,000 rpm. Binding kinetics were determined based on Savitzky-Golay filtering and a 1:1 Langmuir binding model using the respective negative control.

LC-MS
Freshly prepared antibodies and mTG were used for MS ana lysis. LC-MS separations of intact proteins were performed on a TOF 6224A instrument coupled to an HPLC 1260 Infinity, both from Agilent Technologies. The chromatographic column was an Aeris widepore XB-C8, 3.6 µm, 4.6 x 100 mm column from Phenomenex. Eluents consisted of 0.1% formic acid in water (eluent A) and acetonitrile (eluent B). Elution was performed at 0.4 mL/min with a gradient from 10% B to 70% B over 9 min and a total run time of 15 min. The electrospray ionization source was used in positive ion mode and mass spectra were acquired from m/z 100 to 3200. Agilent Mass Hunter software was used for instrument control, data acquisition and analysis (Bioconfirm module).