Mechanistic Investigation of tert-Butanol’s Impact on Biopharmaceutical Formulations: When Experiments Meet Molecular Dynamics

The use of tert-butyl alcohol for the lyophilization of pharmaceuticals has seen an uptick over the past years. Its advantages include increased solubility of hydrophobic drugs, enhanced product stability, shorter reconstitution time, and decreased processing time. While the mechanisms of protein stabilization exerted by cryo- and lyo-protectants are well known when water is the solvent of choice, little is known for organic solvents. This work investigates the interactions between two model proteins, namely, lactate dehydrogenase and myoglobin, and various excipients (mannitol, sucrose, 2-hydroxypropyl-β-cyclodextrin and Tween 80) in the presence of tert-butyl alcohol. We thermally characterized mixtures of these components by differential scanning calorimetry and freeze-drying microscopy. We also spectroscopically evaluated the protein recovery after freezing and freeze-drying. We additionally performed molecular dynamics simulations to elucidate the interactions in ternary mixtures of the herein-investigated excipients, tert-butyl alcohol and the proteins. Both experiments and simulations revealed that tert-butyl alcohol had a detrimental impact on the recovery of the two investigated proteins, and no combination of excipients yielded a satisfactory recovery when the organic solvent was present within the formulation. Simulations suggested that the denaturing effect of tert-butyl alcohol was related to its propensity to accumulate in the proximity of the peptide surface, especially near positively charged residues.


XRD analysis of mannitol's polymorphic states
The crystallization of mannitol is influenced by the manufacturing conditions and has been discussed in the literature [1][2][3] . In fact, the various mannitol anhydrous polymorphs (α, β, and δ) show different thermodynamic stability, which can have an impact on the final features of the active pharmaceutical ingredients. The thermodynamically stable form of mannitol is the β polymorph, while the α and δ polymorphs are metastable 1 . Thus, controlling the crystallization of mannitol is fundamental to guarantee the stability of lyophilized drugs. For this reason, we investigated the effect of TBA and Tween 80 on mannitol polymorphs' formation. Table S1 shows the results of the XRD characterization for the lyophilized mannitol formulations, as listed in Table 1, produced following the freeze-drying protocol described in Table 2 (see manuscript), although without any proteins.
Formation of δ-mannitol with traces of α-and β-for Mo and Mo * formulations was observed, while for formulations M and M * both δ-and β-forms were detected. The corresponding XRD patterns are reported in Figures S1 and S2. The addition of TBA to mannitol formulations promoted the formation of the δ-form, in agreement with reports in literature 4,5 .  Figure S1. XRD pattern of freeze-dried mannitol (M, as indicated in Table 1 of the manuscript) and mannitol's anhydrous polymorphs reference spectra (α, β and δ). Figure S2. XRD pattern of freeze-dried mannitol formulations as indicated in Table 1 of the manuscript.

Development of a TBA Force Field
The Kirkwood Buff Integrals 6 (KBIs, see Eq. 1 of the manuscript) computed for both the original and modified van der Vegt parameterizations of TBA (sims. type 1 of Table 3, see manuscript)  whereas the CHARMM TIP3P 10 /combination rule 2 description was inaccurate at low concentrations, which are, unfortunately, the range of interest of this study. Similarly, Figure S3b shows the TBAwater interactions, evaluated in terms of the TBA-water Kirkwood-Buff integral GTW, which were accurately described by the SPCE/combination rule 1 but rather poorly by the CHARMM TIP3P/combination rule 2 parameterization. Instead, Figure S3c shows that the two TBAff descriptions for the water-water interactions, evaluated in terms of the water-water Kirkwood-Buff integral GWW, show fluctuations.
Overall, the SPCE/combination rule 1 parametrization approximated the experimental KBIs quite well, while the CHARMM TIP3P/combination rule 2 parameterization was inaccurate at low concentrations. The simulations of the same systems with larger boxes (8 nm edge) confirmed the results observed for the smaller systems. Despite the shortcomings of the CHARMM TIP3P/combination rule 2 parameterization, we presented its results as they could still provide valuable qualitative information concerning the TBA behavior and a term for comparison. Figure S3. KBIs for the TBA-water mixtures corresponding to simulations 1a-f (see Table 3 of the manuscript). GTT, GTW, and GWW refer to TBA-TBA, TBA-water, and water-water interactions, respectively.
The 'EXP' label refers to experimental data from Nishikawa et al. 11 , 'Lee and van der Vegt' indicates the original Lee and van der Vegt description 8 , while the 'CHARMM TIP3P_cr2' and 'SPCE_cr1' labels indicate the data obtained from our modified van der Vegt parameterizations. The labels 'S' and 'L' refer to the systems of size 4x4x4 nm and 8x8x8 nm, respectively.

Myoglobin setup
The molecular structure of myoglobin was obtained from the RCSB PDB data bank 12 , PDB code 1WLA 13 . The protein conformation at pH 3.7 was then obtained from the H++ server 14 , resulting in a charge equal to +10, which was balanced by the addition of a corresponding number of Clcounterions. Mb was described using both the CHARMM36m 15 force field in combination with CHARMM TIP3P water and the GROMOS 54a7 [16][17][18] force field with SPCE water. To assess if the resulting structures were equilibrated, a simulation in water was performed, as detailed in the corresponding section of the manuscript (Simulation of Myoglobin Formulations).
The time profiles of RMSD (root mean square deviation) were evaluated for both topologies. The RMSD profiles for Mb's peptide backbone were computed with the gmx rms built-in Gromacs command. Specifically, the reference structure was the structure resulting from the removal of the heteroatoms and the charge adjustment to reproduce the conformation at pH 3.7. Such profiles are reported in Figure S4. Thus, it could be observed that the structures were equilibrated throughout the simulation, as the RMSD fluctuations over time were limited after the initial 20 ns.

RMSD time profiles for myoglobin's formulations
To verify that the simulation setup was adequate to observe the onset of any unfolding/denaturation behavior, as well as long enough to obtain equilibrated configurations, the RMSD time profiles were computed for the peptide backbone of Mb. To do so, firstly the simulations of Mb in bulk water without excipients were used to extract the reference structure for the computation of the RMSD profiles. Specifically, the gmx cluster built-in Gromacs command was used, using the Daura algorithm 19 . The mean structure of the prevalent structure cluster was extracted and hence used as reference. The procedure was performed for both force field descriptions of Mb. It was then possible to compute the RMSD time profiles with such structures as reference. An example of such profiles is reported in Figure S5, where the RMSD time profiles were reported for systems containing Mb without excipients (corresponding to sims. 2a, 3a, 4a, 5a of Table 3 of the manuscript).  Table 3 of the manuscript.
Overall, the trend of the RMSD values is in agreement with the radius of gyration ones, presented in the following section and discussed in the manuscript (see the 'TBA Denatures Myoglobin in Molecular Dynamics Simulations' section). Thus, it could be concluded that the onset of denaturation was observed throughout the simulations, as the values of RMSD were higher whenever TBA and/or interfaces were present, and that the observed conformational state of Mb was equilibrated.

Radius of gyration values for myoglobin's simulations
In this section, the time-averaged values of the radius of gyration for all myoglobin's simulations are reported. The time average was computed over the last 80 ns of the simulations.   Density profiles for myoglobin's formulations at the air-water interface Figure S7. Dimensionless density profiles for excipients and TBA at the A/W interface, corresponding to simulations from 4b to 5d of   blue, positively charged residues; red, negatively charged residues; green, polar residues; yellow, non-polar residues. The labels refer to the corresponding type of simulation (see Table 3 of the manuscript). The first two rows refer to results obtained with the CHARMM36m 15 description of myoglobin, while the last two rows refer to the GROMOS 54a7 [16][17][18] description. Snapshots were realized with Visual Molecular Dynamics 1.9.3 (VMD) 20 .