EGFR binding Fc domain-drug conjugates: stable and highly potent cytotoxic molecules mediate selective cell killing

Preparation of antibody fragments

Amino acid sequences of Fcabs were taken from literature (Tuna et al. 2018). Sequences of Fcabs (no core hinge region, D265A) are given along with huFc (D265A) and Cetuximab sequences (SrtA tag) in the SI. Encoding sequences were ordered as codon-optimized versions and cloned into pTT5 vector for mammalian expression (GeneArt, Thermo Fisher Scientific). Fcabs and huFc controls were expressed by transient transfection of Expi293F^™ cells following the manufacturer’s instructions and the supernatants were harvested after five days post transfection. C-Fab contained a His₆-Tag for purification and was dialyzed against phosphate-buffered saline (PBS) pH 7.4 before purification by immobilized metal affinity chromatography (1 mL HisTrap^™ HP, GE Healthcare) using an ÄKTA Pure device (GE Healthcare). Fcabs, huFc and C-IgG were purified by protein A affinity chromatography using HiTrap^™ Mab Select SuRe 5 mL columns (GE Healthcare) and subsequently formulated in PBS pH 6.8 using HiPrep^™ 26/10 desalting columns. Antibody purity was analyzed by analytical size exclusion (SE) HPLC using a TSKgel^® SuperSW3000 column (Tosoh Bioscience) and by SDS gel electrophoresis. Identity of proteins was confirmed via intact mass analysis by LC-MS using an Exion LC and TripleTOF^® 6600+ mass spectrometer (AB Sciex). Proteins were concentrated using Ultra centrifugal filter units (3 K MWCO, Amicon^®), sterile filtered and protein concentration was determined by UV–VIS spectroscopy at 280 nm. Proteins were snap-frozen in liquid nitrogen and stored at −80 °C.

Preparation of MMAE conjugates

Fcabs and huFc were conjugated to drug-linker Gly₃-Val-Cit-PAB-MMAE (1, Levena) using a genetically engineered mTG (Dickgiesser et al. 2020). mTG-mediated antibody conjugation was performed using 5 mg/mL Fcabs/huFc, 20 molar equivalents of drug-linker and 60 U/mL mTG in PBS pH 6.8 with up to 10% DMSO. Reaction mixes were incubated at 37 °C for 18 h with gentle shaking, chilled to 10 °C and purified by preparative size exclusion chromatography (SEC).

For SrtA conjugation, C-IgG or C-Fab (5 mg/mL) were formulated in 150 mM NaCl, 50 mM Tris-HCl, 5 mM CaCl₂ pH 7.5. SrtA (Chen et al. 2011) was added to a final concentration of 13 µM along with 10 equivalents of Gly₃-Val-Cit-PAB-MMAE (1) per SrtA recognition motif. The reaction mixture was incubated for 90 min at 25 °C, stopped by the addition of EDTA (final 10 mM) and purified by preparative SEC.

Preparative SEC was performed using either a Superdex^™ 200 Increase 10/300 GL, Superdex^™ 75 10/30 GL or a Superdex^™ 200 prep grade 16/60 column in a 1260 liquid chromatography system (Agilent Technologies) or an ÄKTA Avant device (GE Healthcare) with PBS pH 6.8 as running buffer. Purified conjugates were concentrated using Ultra centrifugal filter units (10 K MWCO, Amicon^®), sterile filtered and protein concentration was determined by UV–VIS spectroscopy at 280 nm. The purified conjugates were subjected to analysis by SE-HPLC and DAR determination (reversed phase (RP) HPLC, LC-MS) as described elsewhere (Dickgiesser et al. 2020), snap-frozen in liquid nitrogen and stored at −80 °C.

Preparation of pHAb-dye conjugates

Fcabs, huFc and Cetuximab controls were formulated in 10 mM sodium-bicarbonate buffer pH 8.5. pHAb amine reactive dye (10 mg/mL 1:1 (v/v) DMSO/H₂O, Promega) was either added at a 2:1 molar ratio (pHAb:antibody) (Fcab-1, Fcab-2, Fcab-3, C-IgG, C-Fab) or a 10:1 molar ratio (huFc), followed by incubation at 25 °C, 450 rpm for 1 h in the absence of light. Excess dye was removed by Dulbecco’s phosphate buffered saline (DPBS) equilibrated Zeba^™ Spin desalting columns (ThermoFisher Scientific) according to the manufacturer’s instructions. Aggregation of pHAb-dye conjugates and fluorescence degree of labeling (DOL^F) were determined by an SE-HPLC method described in the SI.

Peptide mapping

Fcab-1 and Fcab-1-MMAE were deglycosylated with GlycINATOR (Genovis) according to the instruction manual. Deglycosylated molecules were then reduced with 10 mM dithiothreitol (DTT) for 30 min at 56 °C and alkylated with 55 mM iodoacetamide for 30 min at room temperature in the dark. Ten microgram protein was digested with 0.5 µg trypsin (mass spectrometry grade, Promega) at 37 °C overnight.

LC-MS analysis was performed using an Exion HPLC system coupled to a TripleTOF 6600+ mass spectrometer (Sciex). 7.5 µg peptide solution was loaded onto an Aeris PEPTIDE XB-C18 column (Phenomenex, part no. 00F-4506-AN) and eluted with a linear gradient from 5 to 50% buffer B (acetonitrile, 0.1% formic acid; buffer A: water, 0.1% formic acid) within 49 min. Data were acquired with positive polarity and in a TOF-MS mass range from 350 to 2500 m/z and a TOF-MS/MS mass range from 50 to 2500 m/z. Other instrument settings were as follows: ion spray voltage 5.5 kV, source temperature 450 °C, accumulation time 0.25 s for TOF-MS and 0.08 s for TOF-MS/MS, gas1 45 psi, gas2 45 psi, curtain gas 35 psi, declustering potential 80 V, and collision energy was set to dynamic. Data were processed with Genedata Expressionist.

Cell culture

Human cancer cell lines were obtained from the American Type Culture Collection (EGFR positive: MDA-MB-468, A431; EGFR negative: MCF-7) and maintained according to standard culture conditions (37 °C, 5% CO₂, 95% humidity). A431 and MCF-7 cells were cultured in Dulbecco’s modified Eagle’s (DMEM) high glucose medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine and 1 mM sodium pyruvate. MDA-MBA-468 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% FBS, 2 mM L-glutamine and 1 mM sodium pyruvate. For sub-culturing, adherent grown cells were detached by adding 0.05% trypsin-EDTA, diluted with fresh medium and transferred into a new culturing flask.

Cellular uptake

Cells were centrifuged at 500 × g for 5 min, the supernatant was discarded, and cells were resuspended in the respective medium without phenol red at 300,000 cells/mL. The cell suspension (40 µL/well) was seeded into a black 384 clear bottom plate followed by incubation (37 °C, 5% CO₂) in a humid chamber overnight. pHAb-dye labeled proteins were supplemented with 0.3% Tween-20 (final), diluted to 3 µM and added to the cells in triplicates (final 100 nM) using a D300e digital dispenser (Tecan). The cells were immediately transferred to a Cytation 5 cell imaging reader (BioTek) equipped with DAPI and RFP filter cubes and a BioSpa 8 automated incubator (BioTek). Brightfield (objective: 10×, LED intensity: 10, integration time: 13 ms, camera gain: 24) and RFP channel images (ex.: 531 nm, em.: 593 nm, LED intensity: 10, integration time: 50 ms, camera gain: 24) were taken every 2 h over a period of 26 h. About 30 min before the 26 h measurement, the plate was removed from the BioSpa 8 device and 1 μg/mL Hoechst 33,342 (ThermoFisher Scientific) was added via a Tecan D300e digital dispenser for an additional 26 h endpoint DAPI image. Images were processed by the BioTek gen5 data analysis software. The sum of the integrated pHAb dye fluorescence intensities of each image was normalized to the number of cells determined in the DAPI channel and subtracted by the sum of the integrated RFP signal at 0 h (background signal). The cell number and background normalized intensities were divided by the pHAb-dye DOL^F of each construct and plotted against the time. Normalized data was fitted by linear regression in GraphPad Prism (GraphPad Software, Inc.) and intracellular accumulation rates (slopes) were derived. Finally, the relative intracellular accumulation was calculated for each construct with respect to the highest intracellular accumulation rate (here, C-IgG-pHAb on MDA-MB-468 was set 100%).

FcRn and EGFR binding

Kinetic parameters of Fcabs, Cetuximab variants and their respective MMAE conjugates were determined by BLI using the Octet^® RED96 system (FortéBio, Pall) at 30 °C and 1000 rpm agitation speed.

For EGFR binding analysis, Fcab variants (10 μg/mL in DPBS), C-IgG (2.5 μg/mL in DPBS) and respective MMAE conjugates were loaded onto anti-human IgG Fc capture biosensors (AHC) for 60–180 s. C-Fab (2.5 μg/mL in DPBS) was loaded onto anti-human Fab-CH1 second Generation biosensors (FAB2G) for 180 s. Biosensors were then transferred into kinetics buffer (DPBS pH 7.4, 0.02% Tween-20 and 0.1% bovine serum albumin) and incubated for 60 s followed by an association step to EGFR-His₆ (produced in-house). EGFR-His₆ was serially diluted in kinetics buffer in a concentration range varying from 20 to 0.313 nM. Association was monitored for 180, 240 or 300 s followed by a dissociation step in kinetics buffer for 600 s to determine k _on and k _off values. EGFR-His₆ was replaced by kinetics buffer, serving as a negative control and reference. Respective non-binding huFc was used as negative control in each experiment. The buffer reference measurement (control curve) was subtracted from antibody measurements for data fitting and kinetics parameter were determined by using FortéBio data analysis software 12.0 applying a 1:1 global full-fit binding model after Savitzky-Golay filtering.

The FcRn binding assay was performed as described elsewhere (Jäger et al. 2021).

Serum stability

The serum stability assay was conducted as previously described (Dickgiesser et al. 2020) applying some minor modifications: MMAE conjugates were incubated at a final concentrations of 5 µM conjugated MMAE (considering the DAR of each construct) in human and mouse serum. Moreover, serum samples were supplemented with 5 µM deuterated D8-MMAE internal standard prior to 96 h serum incubation.

Cell proliferation assay

For the evaluation of C-IgG-, C-Fab- and Fcab-MMAE conjugates and related compounds, 40 µL of viable cell suspension were seeded into opaque 384-well plates (MDA-MB-468: 2500 cells/well, A431: 9000 cells/well, MCF-7: 5000 cells/well) followed by incubation (37 °C, 5% CO₂) in a humid chamber overnight. Test compounds were added using a D300e digital dispenser (Tecan). Free MMAE, and protein/protein-conjugate solutions were supplemented with 0.3% Tween-20 (final) and diluted to 6 µM (MMAE) or 10 µM (proteins). All wells were normalized to the maximum amount of Tween-20 added. Cell viability was determined after four days using Cell Titer Glo reagent (Promega) according to the manufacturer’s instructions. Luminescence values were normalized to luminescence of non-treated cells and dose-response was fitted using the asymmetric (five parameter) fitting function of GraphPad Prism (GraphPad Software, Inc.) to derive IC₅₀ values.

Preparation of Fcabs, controls and pHAb-dye labeled constructs

As a starting point, we selected three different EGFR-binding Fcabs (Fcab-1, Fcab-2, Fcab-3) from literature. For all three, single-digit nanomolar binding affinities to EGFR have been described (K _D 0.7–2.6 nM) (Tuna et al. 2018). Fcabs were modified by removing the core hinge region (TCPPCP). As a negative control, we included an unmodified human Fc (huFc) fragment. As EGFR-binding references, a Cetuximab-based full length IgG (C-IgG) and a Cetuximab-derived Fab (C-Fab) fragment, both equipped with a sortase A (SrtA) recognition motif (LPETG) at the C-terminal end of the light chain, were included. Fcabs and huFc were expressed with a D265A mutation (Baudino et al. 2008; Shields et al. 2001) to avoid Fcγ receptor (I, II, III) mediated cytotoxicity (Mahalingaiah et al. 2019) and purified by affinity chromatography (Figure S1). Mass spectrometry analysis (LC-MS, Figure S2) confirmed the identity of all molecules. Differential scanning fluorimetry showed lower unfolding transition midpoints (T _m ) for Fcabs (e.g. T _m,1 59 vs. 66 °C; Figure S3) indicating diminished but still acceptable thermal stability of Fcabs compared to huFc. Moreover, we evaluated functional and selective cellular binding of the purified constructs (Figures S4 and S5). For cellular uptake studies all constructs were labeled at random lysines via amine-coupling with a fluorescent dye (pHAb-dye) (Figure 1C) that exhibits very low fluorescence at the neutral pH outside cells but strongly increased fluorescence at the acidic pH in endosomal and lysosomal compartments (Figure S6) (Nath et al. 2016). Successful pHAb-dye labeling and the absence of aggregates were confirmed by analytical size exclusion chromatography (SE-HPLC) (Figure S8). To ensure comparability between individual fluorescence intensities resulting from differently labeled constructs, the fluorescence signals were normalized. A comprehensive description of the underlying experimental procedures can be found in the Supporting Information (SI).

Cellular uptake studies

To assess the general suitability of EGFR-binding Fcabs for intracellular drug delivery, their uptake and accumulation into cancer cells were studied. Therefore, pHAb-dye labeled constructs were incubated on EGFR overexpressing (MDA-MB-468, A431) and EGFR negative (MCF-7) adherent cells and fluorescence was continuously measured over 26 h (Figure S9A). Fluorescence intensities were normalized to account for individual cell numbers and fluorescence intensities of labelled constructs. These intensities were plotted against the time to determine intracellular accumulation rates by linear regression (Figure S9B). All accumulation rates were expressed relative to the rate of C-IgG-pHAb (C-IgG-pHAb on MDA-MB-468) which showed the fastest uptake (Figure 2). Differences in accumulation between MDA-MB-468 and A431 cells directly correlated with the differences in specific EGFR expression density. Remarkably, all Fcabs underwent selective, EGFR-mediated cellular uptake. Interestingly, bivalent and monovalent controls C-IgG-pHAb and C-Fab-pHAb exhibited most pronounced internalization, which may arise from different epitopes addressed by Cetuximab than by the Fcabs (Tuna et al. 2018). To measure the therapeutic threshold (IC₅₀) for respective Fcab-drug conjugates, Fcabs were conjugated to cytotoxic payloads and tested for their cell killing activity.

Figure 2:

Intracellular accumulation of Fcab-, huFc-, C-Fab- and C-IgG-pHAb dye conjugates relative to the accumulation rate of C-IgG-pHAb on MDA-MB-468 cells (100%).

Intracellular accumulation rates were derived from 0 to 26 h incubation at 100 nM and normalized to cell number and individual pHAb-dye fluorescence of each construct. Relative EGFR expression profiles (MDA-MB-468 > A431 > MCF-7) are indicated below bar graph. Error bars show the standard deviation of triplicates.

Generation of Fcab-drug conjugates

Fcab-1, Fcab-2, Fcab-3 and huFc were conjugated by an engineered microbial transglutaminase (mTG) (Dickgiesser et al. 2020) targeting Q295 with Val-Cit-PAB-MMAE possessing a triple glycine handle (1) (Figure 3) (Table 1) and purified by preparative size exclusion chromatography (Figure S10). C-Fab and C-IgG were conjugated to 1 by SrtA (Chen et al. 2011; Gébleux et al. 2019) reaching DARs ranging DAR 0.8–1.1. Analytical SE-HPLC of purified conjugates showed the absence of high molecular weight species (Figure S11A) (Table 1). Surprisingly, analyses via reversed phase chromatography (RP-HPLC) (Figure S11B), hydrophobic interaction chromatography (HI-HPLC) (Table S2) and LC-MS (Figure S12) revealed that all three Fcabs showed an elevated DAR between 2.7 and 2.9. This was not observed for huFc (DAR 2.0) indicating that substrate 1 had been coupled to additional residues in the Fcab scaffold only (Table 1). The additional conjugation sites could subsequently be identified via LC-MS peptide mapping as Q311 and Q438 in the constant C_H2 and C_H3 region (Figures 3A and 4). The disappearance of peaks assigned to unconjugated peptides containing Q295 in the digested Fcab-1-MMAE mixture (Figure 4; peak a, b) indicated that Fcabs were nearly completely conjugated at position Q295 while only partially conjugated at position Q311 and Q438 (total 18–23% conjugated based on DAR 2.7–2.9). As previous mTG conjugation to the same constant region in HER2 Fcabs or native IgGs did not lead to Q311 or Q438 conjugation (Dickgiesser et al. 2020; Jäger et al. 2021), accessibility for mTG and altered conjugation has most likely been driven by structural changes or electrostatic effects in adjacent regions. These changes must be induced by the EGFR paratope inserted to the C_H3 region or by the missing core hinge region of the EGFR Fcabs. All three conjugation positions, namely Q295, Q311 and Q438 lie within the solvent exposed exterior of the Fcab. Consequently, conjugation could theoretically affect accessibility for serum proteases and interfere with FcRn and EGFR binding. To assess this notion, Fcab-1-MMAE, Fcab-2-MMAE and Fcab-3-MMAE were subsequently tested for FcRn and EGFR binding and serum stability.

Figure 3:

Fcab-drug conjugates.

(A) Representative figure of a human Fc portion (PDB ID 5VGP) showing the EGFR binding site located in the C_H3 region as well as the conjugation sites Q295, Q311 and Q438 (EU numbering). (B) Structure of linker-payload 1. (C) Schematic representation of MMAE conjugation.

Figure 4:

Conjugation site identification.

LC-MS chromatogram of digested Fcab-1-MMAE shows conjugated peptides (e–i, k–m) that were not detected in the Fcab-1 preparation. Conjugated peptides eluted at higher retention times due to hydrophobic MMAE. Matched pairs of unconjugated peptides (a–d) eluted at lower retention times and peaks disappeared (a, b) or showed reduced intensity (c) in the Fcab-1-MMAE preparation compared to Fcab-1.

Receptor binding properties of Fcab-drug conjugates

Fcab-MMAE conjugates were analyzed along with controls and non-conjugated parent molecules for their binding affinity to the target receptor EGFR and half-life extending FcRn (Table 2) (Figure S13) (Tables S3 and S4). Biolayer interferometry (BLI) measurements did not reveal impaired binding (K _D ) of conjugates to EGFR and FcRn with differences in K _D beyond the assay variability (twofold) suggesting that attached payloads at positions Q295 do not impact binding functionalities of both receptors. However, for Q311 and Q438 it should be noted that effects on EGFR and FcRn binding cannot be assessed robustly with our BLI assay because of just minor conjugation (18–23%).

Serum stability of Fcab-drug-conjugates

Several studies have shown, that conjugate pharmacokinetics can strongly be influenced by premature cleavage of the linker that connects the protein with the cytotoxic drug (Dorywalska et al. 2015; Shen et al. 2012; Strop et al. 2013). The Val-Cit linker motif is especially prone to cleavage by a carboxylesterase (mCes1c) that is present in mouse serum but absent in human serum (Dorywalska et al. 2016). The extend of this instability heavily depends on the chosen conjugation sites. When Fcab-MMAE conjugates were incubated in mouse and human serum for 96 h, no free MMAE could be detected for all constructs. This indicates that the Val-Cit linker motif is not accessible for mCes1c neither at position Q295 nor at the novel positions Q311 and Q438, hence all positions protect the conjugate from being cleaved prematurely (Table 2).

In vitro cytotoxicity of Fcab-drug conjugates

Next, we evaluated selective cell killing capabilities of the Fcab-drug conjugates in an in vitro cell proliferation assay (Figure 5). All Fcab-drug conjugates showed similar sub-nanomolar inhibitory activity on EGFR positive MDA-MB-468 and A431 cells (IC₅₀ 0.18–0.22 nM and 0.23–0.32 nM, respectively) while toxicity against EGFR negative MCF-7 cells was decreased by several orders of magnitude (IC₅₀ > 100 nM) indicating strong target-dependent cell killing. Non-targeting huFc-MMAE showed low toxicity as well (MDA-MB-468: IC₅₀ > 300 nM; A431 and MCF-7: IC₅₀ > 100 nM) confirming that Fcab-MMAE toxicity is primarily driven by specific receptor-mediated uptake. In line with higher cellular uptake on MDA-MB-468 cells compared to A431 cells (Figure 2), Fcab-drug conjugates and Cetuximab controls showed higher activity on MDA-MB-468 cells compared to A431 (Figure 5B). Furthermore, higher potency of Fcab-drug conjugates on EGFR positive cells (IC₅₀ 0.18–0.32 nM) compared to C-Fab-MMAE (IC₅₀ 0.78–0.99 nM) or C-IgG-MMAE (IC₅₀ 0.44–0.50 nM) suggested that reduced intracellular accumulation of Fcabs (Figure 2) may be compensated by their higher DAR enabled by additional conjugation to Q311 and Q438 (2.7–2.9 vs. 0.8 and 1.1, respectively). Overall, the results indicate that in vitro potency of Fcab-drug conjugates is in the typical sub-nanomolar range of ADCs despite their monovalent binding mode.

Figure 5:

Cell proliferation assays.

(A) Cell viability on EGFR positive (MDA-MB-468, A431) and EGFR negative cells (MCF-7). Cells were incubated with serial dilution of MMAE conjugates and free MMAE for 4 days before cell viability was analyzed. Error bars represent standard deviation (SD) of triplicates (B) Inhibitory activity of MMAE conjugates and free MMAE. IC₅₀ values are given as mean of three independent experiments (IC₅₀ ± SD).