Assessment of Monoclonal Antibody/Fc Receptor Interactions Using Bio-Layer Interferometry: Establishment of a Toolbox Panel for Characterization of Therapeutic mAbs and mAb Biosimilars

Dan Papa, Tiffany Walker and Michael Sadick*, Catalent Pharma Solutions, Large Molecule Analytical Chemistry, Kansas City, MO

Characterization of originator and/or biosimilar IgG monoclonal antibody (mAb) therapeutic molecules requires, among other orthogonal assessments, measurements of the interaction of the Fc region of the mAb with all of the potential human IgG-Fc-binding cell-surface receptors.

BLI Basics

FIGURE 1: Diagram of BLI activity with immobilized binding partner attached.

This list includes the low affinity CD16a and CD16b receptors (FcγRIIIA and FcγRIIIB, respectively), which are responsible for the majority of effector function — antibody-dependent cellular cytotoxicity, or ADCC — associated with mAb therapeutics (chiefly IgG1). Also included are the low affinity receptors, CD32a and CD32b/c (FcγRIIA and FcγRIIB/C, respectively), as well as the high affinity receptor CD64 (FcγRI) which are also responsible for mediating multiple immune responses. Completing the list of human receptors for the Fc region of IgG mAbs is the 'neonatal' FcRN, which has great impact on the in vivo half-life of a therapeutic mAb.

It is important to test every therapeutic mAb for binding with the complete panel of human IgG-Fc-binding receptors for characterization of original therapeutics and/or testing of biosimilar mAbs. If the mAb is an effector mAb, then it is necessary to show consistency in binding to the high and low affinity receptors CD64, CD32 and CD16. If the mAb is a neutralizing moiety only (no effector function), one would then expect minimal binding, if any, to the Fcγ receptors. However, all antibodies will bind in some fashion to the FcRN, which despite the name 'neonatal' is present in adult bone marrow-derived cells, as well as other tissue. When an antibody binds to the FcRN, the Ab/FcRN complex is endocytosed and ends up in an endosomal vesicle with a low pH. In the low pH environment of the endosome, the FcRN is in its high affinity state and as such retains a very tight hold on the antibody, essentially sequestering it from the blood stream. When the endosome recycles back to the cell surface, the Ab/FcRN complex is exposed to neutral pH serum/plasma, and the FcRN shifts to its low affinity state, allowing the Ab to detach. It is in this way that the FcRN is thought to play an important role in the PK of mAb therapeutics.

With the preceding ideas in mind, we at Catalent Pharma Solutions, Large Molecule Analytical Chemistry division in Kansas City, MO, set out to essentially develop a tool-box approach for assessing mAb/Fc receptor binding kinetics. The intent was to eventually use the analytical panel both for characterization of biosimilar molecules and comparison to originator molecules, as well as for CMC-related release and stability testing. While use of surface plasmon resonance (SPR) was one possible approach, we chose to proceed with the more robust, user-friendly and more economical method of Bio-Layer Interferometry (BLI; Octet® RED96 system). To that end, a kinetics binding panel was developed on the Octet RED96 platform with recombinant CD16a (FcγRIIIA), CD16b (FcγRIIIB), CD32a (FcγRIIA), CD32b/c (FcγRIIB/C), CD64 (FcγRI) and FcRN. While it's clear that for each and every different mAb or mAb derivative therapeutic molecule the assay panel will need to be optimized, at the least a common and proven starting point can be established.

BLI is an optical analytical technique that analyzes the interference pattern of white light reflected from two surfaces: a layer of immobilized protein on the biosensor tip, and an internal referencev layer. Any change in the number of molecules bound to the biosensor tip causes a shift in the interference pattern of the reflected light. Molecular interactions are measured in real time, providing the ability to monitor binding specificity, rates of association and dissociation.

Similar to SPR techniques, there exist multiple strategies to immobilize the initial binding partner to the biosensor tip, including reactive amine-based binding, biotin/streptavidin binding, Protein A, etc. The Octet system can be used both for quantitative binding assays (similar to ELISA) and for binding kinetic assays (KON/KDIS/KD determinations; KD = KON/KDIS).

Our strategy at Catalent was based upon the use of the Octet RED96 system to determine binding kinetics of Fc receptors to monoclonal antibodies. The common component for assaying all potential therapeutic molecules was to be the Fc receptors, while the actual test samples would be the therapeutic mAbs as characterization, release, and stability test samples. The following strategy was adopted from the outset: the Fc receptors (each type tested separately) would be immobilized to the Octet biosensor tips, and the 'target' mAbs would be in solution at various concentrations for association/dissociation testing. This strategy would streamline our workflow, allowing us to avoid any derivatization of the test analyte (the mAbs), and more objectively assess the state of the particular mAb for binding to the receptor panel. To this end, HIS-tagged recombinant receptors FcγRIIIA, FcγRIIIB, FcγRIIA, FcγRIIB/C, FcγRI) and FcRN were obtained from R&D Systems. Ni-NTA biosensors, which strongly bind HIS-tagged molecules, were used for the actual Octet analyses.

For each class as well as subclass of Fc receptor, the following conditions required optimization: receptor concentration, receptor load (immobilization) time, mAb concentration/serial dilution, mAb association time and dissociation time. As each class of receptor has a different affinity for Fc, optimized conditions were potentially different for each type of receptor. For FcRN, two different optimal conditions were determined, one for neutral pH (7.4)/low affinity and one for low pH (6.0)/high affinity.

As a whole, several basic underlying requirements for successful analyses of IgG binding to any of the receptors became clear. First off, it is important to start with as low a concentration of the receptor as is practical to be immobilized on the biosensor tip. Next, providing long immobilization incubations, of six minutes or more was also important, as was providing ample time for IgG/FcR association and dissociation such that adequate curvature in the mAb binding response could occur. An association time of three minutes seemed to be sufficient for all Fc receptor/IgG pairings. A dissociation time of 10 minutes worked out to be the most optimal, again, for all Fc receptor/IgG pairings.

Figure 2 shows the processed data from one of the conditions used for the IgG/FcγIIIA interaction. The four association/dissociation curves shown are the four of seven IgG concentrations that provided the most consistent results for this interaction.

The binding kinetics were modeled using both 2:1 modeling (Figure 3) and 1:1 modeling (Figure 4).

Both 1:1 and 2:1 modeling using the full global fit gave similar values for kon, kdis and KD. Visually, 2:1 modeling appeared to give a better fit (modeled curve versus actual data). Statistical analyses and value derivations from the 2:1 and 1:1 modeling are shown in Tables 1 and 2.

FIGURE 2: Example of one condition (3-minute association, 10-minute dissociation) of FcγRIIIA binding kinetics, Processed Data.

FIGURE 3: Example of one condition of FcγRIIIA binding kinetics, stacked graphs (2:1 modeling).

FIGURE 4: Example of one condition of FcγRIIIA binding kinetics, stacked graphs (1:1 modeling).

Loading
Sample
(Receptor)
Conc. IgG
(nM)
Response KD (M) KD2 KD
Error
KD2
Error
Full
X^2
Full
R2
kon
(1/Ms)
kon2 kon
Error
kon2
Error
kdis(1/s)
0.1 µM 1000 0.1362 5.2E-07 7.4E-07 3.0E-08 3.7E-08 0.0066 0.996 4.4E+03 6.3E+04 2.5E+02 2.8E+03 2.2E-03
0.1 µM 500 0.0769 5.2E-07 7.4E-07 3.0E-08 3.7E-08 0.0066 0.996 4.4E+03 6.3E+04 2.5E+02 2.8E+03 2.2E-03
0.1 µM 250 0.0392 5.2E-07 7.4E-07 3.0E-08 3.7E-08 0.0066 0.996 4.4E+03 6.3E+04 2.5E+02 2.8E+03 2.2E-03
0.1 µM 125 0.0252 5.2E-07 7.4E-07 3.0E-08 3.7E-08 0.0066 0.996 4.4E+03 6.3E+04 2.5E+02 2.8E+03 2.2E-03

Table 1: Example of one condition of FcγRIIIa binding kinetics, 2:1 Modeling Results Table.

Loading
Sample
(Receptor)
Conc. IgG
(nM)
Response KD (M) KD
Error
Full
X^2
Full
R2
kon
(1/Ms)
kon
Error
kdis
(1/s)
kdis
Error
Rmax Rmax
Error
kobs
(1/s)
0.1 µM 1000 0.1362 1.5E-07 3.6E-09 0.0574 0.968 2.7E+04 6.1E+02 4.1E-03 3.8E-05 0.1369 0.001 3.2E-02
0.1 µM 500 0.0761 1.5E-07 3.6E-09 0.0574 0.968 2.7E+04 6.1E+02 4.1E-03 3.8E-05 0.0949 0.0012 1.8E-02
0.1 µM 250 0.0387 1.5E-07 3.6E-09 0.0574 0.968 2.7E+04 6.1E+02 4.1E-03 3.8E-05 0.071 0.0013 1.1E-02
0.1 µM 125 0.0251 1.5E-07 3.6E-09 0.0574 0.968 2.7E+04 6.1E+02 4.1E-03 3.8E-05 0.0709 0.0018 7.5E-03

Table 2: Example of one condition of FcγRIIIa binding kinetics, 1:1 Modeling Results Table.

FcR Type CD Nomenclature Octet-determined Mean KD Values
(M) (5–6 Different Conditions)
Averaged
KD Error
Averaged KD Error
(% of KD)
Octet-determined KD Value Variability
(%CV, n = 5–6)
FcγRI CD64 2.2 x 10-9 8.3 x 10-11 4% 18%
FcγRIIA CD32a 1.4 x 10-7 3.6 x 10-9 3% 11%
FcγRIIB/C CD32b/c 2.4 x 10-7 1.4 x 10-8 6% 37%
FcγRIIIA CD16a 1.8 x 10-7 4.7 x 10-9 3% 30%
FcγRIIIB CD16b 1.0 x 10-7 1.4 x 10-9 1% 23%
FcRN (@pH7.4) NA 2.1 x 10-7 5.1 x 10-9 2% 53%
FcRN (@pH6.0) NA 1.6 x 10-9 4.3 x 10-11 3% 34%

Table 3: Summary of binding kinetics for Fc receptor panel. Results derived from 1:1 Modeling.

While the associated R2 was slightly lower in the 1:1 modeling (0.969 vs. 0.996), agreeing with the visually smoother fit, the 1:1 modeling KD Error was appreciably lower (2% KD error vs. 6% KD error), compared to that derived from 2:1 modeling, indicating that 1:1 modeling was actually the more precise and correct method. Thus the 1:1 modeling derivation of binding kinetics was used to assess all 6 Fc receptors in the panel. A summary of the results for the 6 different Fc receptors is shown in Table 3. The values shown are averaged KD values derived from different conditions for each given Fc receptor of receptor load time, IgG association time and dissociation time. The values shown for Octet-determined KD value variability (%CV) demonstrate a good degree of robustness around the binding conditions. While the %CV values for some of the assays appear fairly high, the data averaged for each antibody-receptor pair in Table 3 was from experiments that used varying assay conditions. With further optimization, as well as a consistent optimized set of binding parameters for each receptor, we are confident the CV's will be below 20% when averaging multiple executions of the same receptor/mAb combination.

The results shown for the KDs of IgG for the different Fc receptors fall in line with expectations. FcγRII and FcγRIII show low to medium binding kinetics in the µM range, whereas FcγRI is high affinity with KD values in the nM range. FcRN has µM range binding at neutral pH (low affinity condition), but nM range binding at acidic pH (high affinity condition). The FcRN

binding kinetics at the two different pH conditions is consistent with the current understanding of the function of the FcRN receptor, and how it impacts antibody half-life in vivo.

In conclusion, we have successfully demonstrated development and establishment of a platform for characterization of IgG/Fc receptor binding at Catalent utilizing a panel of appropriate Fc receptors and easy-to-use Octet instrumentation. As the immobilized moieties for this panel are the Fc receptors, the analyte of interest (the mAb) is in no way derivatized, allowing objective analysis of the receptor panel with any mAb. With minimal additional optimization, the Fc receptor panel established here can be quickly adapted for any mAb. We are in the process of finalizing validation of the Octet RED96 system, and will be further developing the Fc receptor panel testing for use not only in characterization, but in cGMP release and stability assessments when appropriate.

Acknowledgements

The authors would like to express their sincere gratitude to David Apiyo, Ph.D., Field Applications Scientist, Pall ForteBio, for essential guidance with strategy and data analysis/interpretation, and great patience.

*To whom all correspondence should be addressed:

Michael Sadick, Ph.D.
Catalent Pharma Solutions
10245 Hickman Mills Drive
Kansas City, MO 64137
mike.sadick@catalent.com

Catalent