Kinetic mechanism of controlled Fab-arm exchange for the formation of bispecific immunoglobulin G1 antibodies

Bispecific antibodies (bsAbs) combine the antigen specificities of two distinct Abs and demonstrate therapeutic promise based on novel mechanisms of action. Among the many platforms for creating bsAbs, controlled Fab-arm exchange (cFAE) has proven useful based on minimal changes to native Ab structure and the simplicity with which bsAbs can be formed from two parental Abs. Despite a published protocol for cFAE and its widespread use in the pharmaceutical industry, the reaction mechanism has not been determined. Knowledge of the mechanism could lead to improved yields of bsAb at faster rates as well as foster adoption of process control. In this work, a combination of Förster resonance energy transfer (FRET), nonreducing SDS-PAGE, and strategic mutation of the Ab hinge region was employed to identify and characterize the individual steps of cFAE. Fluorescence correlation spectroscopy (FCS) was used to determine the affinity of parental (homodimer) and bispecific (heterodimer) interactions within the CH3 domain, further clarifying the thermodynamic basis for bsAb formation. The result is a clear sequence of events with rate constants that vary with experimental conditions, where dissociation of the K409R parental Ab into half-Ab controls the rate of the reaction.


Contents:
S-2,3: Supplementary materials and methods S-4: Figure S1 showing SDS-PAGE of IgG1WT and IgG1C S Abs S-5: Figure S2 showing SEC of IgG1WT and IgG1C S Abs S-6: Figure S3 showing HIC after cFAE reaction using IgG1WT and IgG1C S Abs S-7: Figure S4 showing cFAE kinetics with two different pairs of IgG1WT Abs S-8: Figure S5 showing cFAE kinetics for IgG1 and IgG4 Abs S-9: Figure S6 showing raw data for cFAE under different reducing conditions S-10: Figure S7 showing raw data for cFAE at different pH S-11: Figure S8 showing raw data for cFAE at different ionic strength S-12: Figure S9 showing cFAE for different K409 mutants S-13: Figure S10 showing alternative measurement of half-Ab KD values S-14: Figure S11 showing the kinetics of hinge redox for parental and bsAbs S-15: Figure S12 showing comparison of half-Ab kd and KD S-16: Figure S13 showing SDS-PAGE of purified IgG1 Fc S-17: Figure S14 showing the effect of dye:Ab ratio on cFAE kinetics S-18: Scheme S1 showing derivation of Keq for cFAE S-19: Scheme S2 showing calculation of %bsAb based on half-Ab KD values S-20,21: Scheme S3 showing derivation of FCS fitting equations S-22: Supplementary references

Kinetics of cFAE measured by FRET.
To determine the kinetics of parental pairs containing the α-CD20 IgG1 parental with alternative K409 substitutions, parental Ab-dye conjugates were mixed in a 1:1 ratio at a final concentration of 167 nM in 18 μL of PBS at 25 ºC in 96-well PCR plates. The FAE reaction was initiated by adding 9 μL of 2-MEA (final concentration 25 mM) and incubating in an iQ™5 Multicolor Real Time PCR Detection System (BioRad) set at 25 ºC. FAE was monitored every 2 minutes by exciting at 494 nm and measuring emission at 620 nm. FRET signal was normalized by subtracting the minimum fluorescence and dividing by the maximum FRET in a dataset. Rates were determined using Sigmoidal dose-response (variable slope) fits in GraphPad Prism 7.

Equilibrium constants (Fluorescence-assisted high-performance liquid chromatography)
To determine equilibrium constants, antibodies and DyLight-488 labeled antibodies were reduced with 10 mM DTT (60 min/37 o C) and alkylated with 22 mM of iodoacetamide. Serial four-fold dilutions of reduced/alkylated IgG (0.003 -1000 nM half-molecules) were incubated with 0.1 ng/ml reduced/alkylated F405L-488 (for the bsAb) or 0.5 ng/ml K409R-488 (for the K409R antibody) in PBS containing as a carrier protein 0.1 mg/ml certolizumab pegol, and incubated at 37 o C for up to 1 day before analysis. Between 50 -1000 μl of a sample was applied using a thermostatted autosampler (20 o C) to a Superdex 200 HR 10/300 column, which was connected to an ÄKTAexplorer HPLC, and eluted at 0.5 ml/min. Elution profiles were monitored by measuring the fluorescence (excitation/emission 488/525 nm) with a Prominence RF-20Axs in-line fluorescence detector (Shimadzu, Kyoto, Japan). To calculate dissociation constants, backgroundcorrected fluorescence intensities (F) corresponding to the peak maxima of either bound or free labeled protein were plotted against the concentration of antibody x (molar concentration of the number of half-molecules), and a homodimerization model (F = F0 + 0.25 × ΔF × [KD + 4x -(KD 2 + 8x × KD) 0.5 ]/x), where F0 is the fluorescence at zero concentration, ΔF is the fluorescence at saturating concentrations minus F0, and KD is the dissociation constant) was fitted to the data using Microcal Origin 7.0 software. For the bsAb, an apparent dissociation constant was measured as F = F0 + ΔF × [(x + y + KD) -((x + y + KD) 2 -4xy) 0.5 ]/2y, with x = y (molar concentrations of halfmolecules of the K409R and F405L variants, resp). Standard errors were calculated from duplicate or triplicate values of the KD as determined in independent experiments.

Equilibrium constants (FRET)
Equilibrium constants of F405L was determined by recording fluorescence spectra using a Nanodrop ND3300 of two-fold dilutions of equimolar mixtures of F405L -488 and F405L -594 after incubation for 2 hours at 37 o C in the presence of 3 mM DTT and subsequent equilibration at 20 o C. Spectra were corrected for background fluorescence and the ratio of F620/F588 was plotted against total concentrations of half-molecules (A+B) to obtain a dose-response curve. The concentration of mixed F405L dimers (AB) will correlate linearly with the total amount of F405L dimers (A2 + B2 + AB), therefore, the FRET signal is representative of the amount of dimers present in solution. The dissociation constant was calculated by fitting a homodimerization model to the data (see above), reported value is the average of 5 different experiments.

Hydrophobic interaction chromatography (HIC)
HIC was used as an orthogonal technique to FRET to ensure formation of >90% bsAb for IgG1WT and IgG1C S Abs containing F405L/K409R mutations. After performing cFAE reactions including dialysis to remove reducing agent, 20 µg of protein was injected onto a butyl-NPR column (Tosoh Bioscience, 42168) and parental Abs were separated from bsAb using a gradient from 1.5 M to 0 M (NH4)2SO4 in 0.1 M sodium phosphate, pH 6.5 at 0.5 mL/min. HIC of purified parental Abs was also performed to identify which protein each peak represents.

Figure S10. Analysis of bsAb interaction strength. A fixed amount of DyLight488-labeled F405L (A and C) or DyLight488-labeled K409R (B) was incubated with different concentrations of unlabeled bsAb (A), K409R (B) or F405L (C) parental (all reduced), respectively, and analyzed with high performance size-exclusion chromatography (left panels). Fluorescence at 14.5 ml is plotted vs concentration of half-molecules (right panels) and analyzed to calculate the KD value.
The F405L dissociated rapidly and with the column assay a peak shift was observed rather than two populations (half-molecules and homo-dimers). The dissociation constant was therefore alternatively measured using a FRET assay (C; middle and right panels): Equimolar amounts of DyLight488 and DyLight594-labeled F405L were incubated at various concentrations in the presence of DTT and fluorescence spectra recorded. Normalized fluorescence is plotted vs concentration of half-molecules and analyzed to calculate the KD value.

Figure S12. Comparison of half-Ab dissociation rates and half-Ab dimerization affinities. Y-axis is the half-Ab dissociation rate, measured by FRET with different pairs of Alexa 488-and Alexa 594-labeled IgG1C S . X-axis is the half-Ab dimerization affinity, measured by FCS with Alexa 488-labeled half-Fc binding to unlabeled half-Ab.
Values of kd and KD were tightly correlated with R 2 = 1. Note that the linear fit appears curved due to the log axes. Figure S13.