Unraveling the Microscopic Mechanism of Molecular Ion Interaction with Monoclonal Antibodies: Impact on Protein Aggregation

Understanding and predicting protein aggregation represents one of the major challenges in accelerating the pharmaceutical development of protein therapeutics. In addition to maintaining the solution pH, buffers influence both monoclonal antibody (mAb) aggregation in solution and the aggregation mechanisms since the latter depend on the protein charge. Molecular-level insight is necessary to understand the relationship between the buffer–mAb interaction and mAb aggregation. Here, we use all-atom molecular dynamics simulations to investigate the interaction of phosphate (Phos) and citrate (Cit) buffer ions with the Fab and Fc domains of mAb COE3. We demonstrate that Phos and Cit ions feature binding mechanisms, with the protein that are very different from those reported previously for histidine (His). These differences are reflected in distinctive ion-protein binding modes and adsorption/desorption kinetics of the buffer molecules from the mAb surface and result in dissimilar effects of these buffer species on mAb aggregation. While His shows significant affinity toward hydrophobic amino acids on the protein surface, Phos and Cit ions preferentially bind to charged amino acids. We also show that Phos and Cit anions provide bridging contacts between basic amino acids in neighboring proteins. The implications of such contacts and their connection to mAb aggregation in therapeutic formulations are discussed.

Table S1: The number of buffer ions, NaCl and water molecules for different systems simulated in this work (see table 1 in the main text).The variability in the number of water molecules stems from different initial position of the buffer ions in the different copies of the systems.The difference is smaller for the Phos systems due to their smaller molecular size.
System System name N P hos − /N P hos 2− N Cit 2− /N Cit 3− N N a + /N Cl − N wat 1. Fab

Calculation of survival probability
We calculated the time-series of the shortest of all atomic-pair distances, d min , between each buffer molecule and the protein form our MD trajctories.d min defines the separation between a buffer molecule and the protein surface.A buffer molecule was deemed to be adsorbed on the protein if d min ≤ 0.4 nm.The cutoff was set such that both hydrogen bonds (cutoff acceptor-hydrogen distance of ∼0.25 nm) and salt-bridges (cutoff distance of ∼0.4 nm) are included (D Xu, CJ Tsai andR Dussinov, Protein Eng. 1997, 10, 999-1012 ).
The time-series of d min was calculated for each buffer molecule.We show in Figure S3   The time dependence of d min for all buffer molecules of a given type and charge state was used to study the adsorption kinetics by calculating the survival probability S(t), Here, h(0) = 1 if a buffer-protein contact is present at time t = 0 and 0 otherwise, while h(t ′ ) is 1 if a contact, present at t=0, is still intact at time t = t ′ .If re-attachment takes place due to diffusion of a given buffer molecule back from the solution, we consider this event as a new adsorption event.S(t) can thus be defined as the probability that a buffer-protein contact that exists at time 0, continues to exist at least up to time t.As S(t) depends on the strength of buffer-protein interaction, a comparison provides information on the relative affinities of different buffer species and charge states towards the protein surface.
For calculating the survival probability of buffer-mediated bridges between protein surfaces, a buffer molecule was said to be bridging the protein surfaces if it had a d min < 0.4 nm for both protein surfaces simultaneously.This criterion was then used for the calculation of residence times and survival probabilities as discussed above.
the variation of d min with time for a single Cit 3− buffer molecule as an example.The trajectory can be decomposed into a series of intervals: time regions where the d min for a buffer molecule either lies within or beyond a distance of r cut from the protein surface.The stretch of time for which d min ≤ r cut corresponds to a residence event and the time interval is called the residence time (τ r ).

Figure S3 :
Figure S3: Time dependence of the minimum distance, d min , of a Cit 3− buffer molecule from the Fc surface.The regions of the trajectory where d min is below r cut = 0.4 nm correspond to adsorption events (shaded in blue) while the rest of the trajectory corresponds to free diffusion of the buffer molecule in solution.The length of an adsorption event τ r defines the residence time of the buffer for that event.

Figure S5 :FigureFigureFigureFigure S10 :
Figure S4: (Top) Compensation of the protein charge as a function of the distance, r, to the protein surface.(Bottom) Fitting of the Fc-cit and Fc-phos to an exponential decay (a • exp(−r/b)).The fitting was performed using the data in the region 1-2.5 nm, where the charge Z ef f (r) varies exponentially.The number in the legend shows the fitting parameters.with b, being the Debye length, ξ.