Antibody PEGylation in bioorthogonal pretargeting with trans-cyclooctene/tetrazine cycloaddition: in vitro and in vivo evaluation in colorectal cancer models

Bioorthogonal chemistry represents a challenging approach in pretargeted radioimmunotherapy (PRIT). We focus here on mAb modifications by grafting an increase amount of trans-cyclooctene (TCO) derivatives (0 to 30 equivalents with respect to mAb) bearing different polyethylene glycol (PEG) linkers between mAb and TCO (i.e. PEG0 (1), PEG4 (2) and PEG12 (3)) and assessing their functionality. We used colorectal xenograft (HT29/Ts29.2) and peritoneal carcinomatosis (A431-CEA-Luc/35A7) as tumor cells/mAbs models and fluorescent tetrazines (TZ). MALDI-TOF MS shows that grafting with 2,3 increases significantly the number of TCO per mAb compared with no PEG. In vitro immunofluorescence showed that Ts29.2 and 35A7 labeling intensity is correlated with the number of TCO when using 1,3 while signals reach a maximum at 10 equivalents when using 2. Under 10 equivalents conditions, the capacity of resulting mAbs-1–3 for antigen recognition is similar when reported per grafted TCO and comparable to mAbs without TCO. In vivo, on both models, pretargeting with mAbs-2,3 followed by TZ injection induced a fluorescent signal two times lower than with mAbs-1. These findings suggest that while PEG linkers allow a better accessibility for TCO grafting, it might decrease the number of reactive TCO. In conclusion, mAb-1 represents the best candidate for PRIT.

Spectroscopic data were consistent with previously reported data for this compound 2 . Crude product 3 was then coupled to mAbs as such without purification for preliminary studies. mAb modifications. Both Ts29.2 and 35A7 mAbs were functionalized by addition of different amounts of TCO in the reaction, namely 0, 5, 10, 15, 20 or 30 equivalents of TCO for 1 equivalent of mAb. The distance between mAb and TCO was also modulated by insertion of PEGylated spacers of various lengths, i.e. mAb .
Reactions were stirred for 30 min in the dark at 4 °C and then purified on Zeba desalting columns (40 kDa MW cut-off, 0.5 mL) (Pierce Zeba TM desalting columns, Thermo Scientific). mAbs-1-3 conjugates were kept in the dark either at -20 °C or 4 °C. All reactions were performed at least in triplicates.
Stability studies were made in vitro, following the procedure described in supplementary methods.
For in vivo experiments, the same grafting protocol was applied on 500 µg of mAb. Final concentration of mAbs-1-3 conjugates was measured using a Multiskan GO microplate spectrophotometer (Fisher Scientific, France). We determined a yield of mAb recovery taking into account the starting quantity of mAb and the final one after desalting columns.
Determination of the number of 1,2 and 3 moieties per mAb. Assessments were made using MALDI-TOF MS analyzes (Voyager DE-Pro mass spectrometer, Sciex, USA). Sinapinic acid (Sigma, France) was diluted in acetonitrile/water (30/70, v/v) with 0.1 % TFA at 10 mg/mL to obtain the matrix solution. A set of three serial dilutions from 1 mg/mL of mAb-1-3 in PBS was prepared by mixing those with the matrix solution (2/1, v/v). Acquisitions were performed in a positive linear mode and 600 shots were averaged for each spectrum. The average mAb-1-3 molecular weight was obtained from the mass of the [M+H] + peak for the dilution set. Calibration settings corresponded to a close external mode using IgG1 (AB Sciex, USA). The number of 1-3 grafted per mAb was quantified using the molecular weight difference between the mAbs-1-3 and the unmodified mAbs, net masses added depending on the modification (about 272, 520 and 872 Dalton for 1, 2 and 3 respectively). mAbs-1-3 stability. Stability of both Ts29.2-1-3 and 35A7-1-3 was assessed in vitro with TZ-5-FAM.
After grafting TCO and PEG-TCO moieties mAbs were aliquoted and either frozen at -20 °C or keep at 4 °C. TCO/TZ interaction was assessed 7, 14 and 28 days after storage. 5 µg of mAbs-1-3 was added to Laemli 4X and heated to 95 °C for 5 min.
Samples were then loaded on SDS-PAGE acrylamide gels 4-15 % (Biorad, France). After migration, gels were entirely incubated 5 min in a solution containing 0.02 mM of TZ-5-FAM (10-14 equivalents with respect to TCO) and then rinsed under gentle shaking 15-20 min in deionized water before being imaging (G:Box, Ozyme, France). Gels were finally colored with Simplyblue™ SafeStain following the manufacturer's instructions (Thermofisher Scientific, France) and imaged with Chemidoc imager (Biorad, France). Quantification was made using both ImageJ and ImageLab software. A ratio between the fluorescence intensity and the amount of protein was made for the main mAb form to assess the reactivity of TCOs. In parallel, aliquots were analyzed by MALDI-TOF MS to determine the mean number of TCO grafted.

In vivo imaging settings
Optical images were acquired using a small animal imaging system (IVIS spectrum, Perkin Elmer, USA) and a dedicated software (Living Image 4.5 software, Perkin Elmer, USA). Acquisitions were performed using the following parameters. Bioluminescence: automatic exposure time; binning: medium; F/stop: 1; excitation filter: blocked; emission filter: open. NIR fluorescence for cyanine 5 imaging: automatic exposure time; binning: medium; F/stop: 2; excitation filter: 640 nm; emission filter: 680 nm. Images were analyzed using Living Image. Regions of interest (ROI) were drawn manually, and light was quantified as photons/seconds/cm 2 /steradian. Signal was represented using inverse rainbow color. Images were processed minimally as no smoothing was applied. All comparative images were threshold at the same MAX and MIN intensities.

Confocal imaging settings
The same settings were applied for acquisitions of all quantified images. We used three lasers: 405 nm for DAPI imaging, 488 nm for FITC and 532 nm for Cyanine3. The same laser intensity was applied for all images in order to compare each mAb. Imaging resolution were 1024 x 1024 pixels. We applied a zoom = 1. An exception was made for co-localization images which are not quantified, images were in 520 x 520 pixels with a zoom of 2. An offset of -0.5 was applied for removing background noise with a gain of 950. Phase correction was equal to -36.92. Bidirectional imaging with Z-steps every 2 µm were made on the whole cell layer. Three random fields per well were imaged and quantified in three or four independent experiments.

Signal quantification on confocal imaging
To quantify the intensity of the signal localized on cell membrane we developed a 3D-automatized method on ImageJ software. Procedure steps is detailed in the following macro:

Supplementary Figures
Supplementary Figure S1: Relation between the number of TCO grafted on 35A7mAbs and their functionality. Number of TCO grafted was determined by MALDI-TOF MS and is expressed as mean values [min-max], n= 3 independent experiments. All IF imaging were made with the same settings.
Yields correspond to the mAb recovery after grafting process. White numbers are mean fluorescence intensity quantified on the corresponding IF imaging. Scale bar: 50 µm.