Characterization of protein-excipient interactions and protein-protein interactions on the molecular level using in-silico simulation methods

The aim of the protein-excipient interactions and protein-protein interactions in formulation (PIPPI) consortium was to systematically investigate the physicochemical behavior of various protein and peptide therapeutics (PPTs), also called biologics. To achieve this, a combination of various biophysical characterization methods and in-silico tools were used.

The overall goal of the PIPPI consortium was to develop an open-access free database, which would contain information about various properties of PPTs in different formulations. My aims, being part of the PIPPI consortium, were to bridge the molecular interactions observed in-silico with the data generated experimentally.

PPTs are often characterized by high specificity and potency with low toxicity and therefore has interested many pharmaceutical industries wanting to develop medicines to treat severe human diseases. Long-term stability is one of the challenges in the formulation of biologics. Currently, a detailed molecular understanding of the effect of different physicochemical formulation conditions on the stability of biologics is sparse as molecular interactions are difficult to probe experimentally at a molecular level.

Computational methods, such as molecular dynamics (MD) simulations, can provide insights about the molecular interactions on the single-molecule level. The PhD thesis describes fundamental research carried out to investigate the effect of pH and varying concentrations of different additives on diverse classes of PPTs.

In the class of peptides, the wild-type plectasin (known as cysteine stabilized antimicrobial defensin) and three variants were investigated using molecular dynamics (MD) simulations in combination with microscale thermophoresis (MST) and nuclear magnetic resonance (NMR). The variants are i) Asn5Ser Asp9Ala Lys26Arg, ii) Asp9Ser Gln14Lys Val36Leu and iii) Asp9Asn Met13Leu Gln14Arg.

Peptide secondary structures stayed intact during the 100ns MD simulations that were carried out at varying pH and ionic strength (NaCl). However, flexibility in the loop containing a distinct anionic tetrapeptide stretch close to the N-terminus increases with pH due to the change in electrostatics.

Based on preferential interaction coefficient calculations, it is observed that sodium ions have a higher preference for the plectasin variants than chloride ions. The conformational stability of the plectasin variants are attributed to the presence of three cystines. Therefore, thermodynamic integration MD simulations supplemented with NMR chemical shift assays were used to determine the order of cystines reduction.

The order of reduction of cystines deduced from NMR results and MD simulations are in good correlation revealing that complete unfolding is only observed upon reduction of all cystines. Peptide-excipient interaction hotspots were deduced from MD simulations in combination with MST and NMR measurements.

Arginine, histidine, Tris, and trehalose showed preferential binding to plectasin. In the classes of proteins, human serum transferrin (Tf) and the conjugate fusion protein human serum albumin–neprilysin (HSA-NEP) were examined. Combining small-angle X-ray scattering (SAXS) and MD simulations, conformational stability of Tf and HSA-NEP in various formulation conditions were studied.

Tf consists of two homologous halves termed as N-lobe and C-lobe, respectively. The SAXS data for Tf showed that Tf prefers to exist in a partially open (PO) conformer (only N-lobe is open) at pH 5, and in the closed form (HO) above pH 5. In-silico studies showed that the main conformational drive from HO to PO conformers at pH below 5 is due to the protonation of Tyr188 and Lys206 that are located in the N-lobe.

At higher pH, both residues are deprotonated and favorably interacts with the iron ion. Therefore, Tf prefers to be in the HO form. Furthermore, it is observed that only at low pH, chloride ions prefer to bind in the iron-binding cleft of the N-lobe. Thus, protonation of Tyr188 and Lys206 together with the binding of chloride ions in the iron-binding site of N-lobe has shown to drive the opening of the N-lobe at low pH.

The addition of different excipients showed that excipient interactions with two specific loop regions can trigger the transformation from HO to PO conformers at low pH. With respect to HSA-NEP, the SAXS experiments performed at varying physicochemical conditions showed that the protein is predominantly present in four different conformational states having a distinct protein-protein interaction interface.

Using in-silico approaches, such as free energy methods and surface electrostatics, it was observed that surface electrostatics is the driving force for the preference of different conformers in varying conditions. Overall, results from SAXS measurements and MD simulations are in good agreement providing a molecular understanding of the behavior of the proteins.

The research findings presented in this thesis indicate the usefulness of combining in-silico and experimental techniques and that this approach can aid in designing new strategies for the formulation of biologics.