Combining Molecular Dynamics Simulations and Biophysical Characterization to Investigate Protein-Specific Excipient Effects on Reteplase during Freeze Drying

Abstract

We performed molecular dynamics simulations of Reteplase in the presence of different excipients to study the stabilizing mechanisms and to identify the role of excipients during freeze drying. To simulate the freeze-drying process, we divided the process into five distinct steps: (i) protein-excipient formulations at room temperature, (ii) the ice-growth process, (iii)-(iv) the partially solvated and fully dried formulations, and (v) the reconstitution. Furthermore, coarse-grained (CG) simulations were employed to explore the protein-aggregation process in the presence of arginine. By using a coarse-grained representation, we could observe the collective behavior and interactions between protein molecules during the aggregation process. The CG simulations revealed that the presence of arginine prevented intermolecular interactions of the catalytic domain of Reteplase, thus reducing the aggregation propensity. This suggests that arginine played a stabilizing role by interacting with protein-specific regions. From the freeze-drying simulations, we could identify several protein-specific events: (i) collapse of the domain structure, (ii) recovery of the drying-induced damages during reconstitution, and (iii) stabilization of the local aggregation-prone region via direct interactions with excipients. Complementary to the simulations, we employed nanoDSF, size-exclusion chromatography, and CD spectroscopy to investigate the effect of the freeze-drying process on the protein structure and stability.

Keywords: Reteplase; arginine; coarse-grained simulations; formulation development; freeze drying; molecular dynamics; protein stability.

Figure 1

Chimeric model of Reteplase. The structures were created by combining 1TPK and 1BDA to model the kringle-2 domain and catalytic domain, respectively. Modeller software was used to build the structure.

Figure 2

The full cycle of the developed FD simulation protocol. The liquid and frozen conditions were simulated in the NPT ensemble. The drying process was simulated in the NVT ensemble. The arrows indicate that the final frame of the previous simulations was used to initiate the next simulations.

Figure 3

The protein–protein interaction (PPI) profile of Reteplase. (a) Electrostatic potential map showing local positively and negatively charged patches. (b) Visual representation of the hydrophobic aggregation-prone residues as red sticks. The aggregation-prone residues were detected using the A3D algorithm [56]. (c,d) Interaction-prone residues observed in the CG simulations. Regions involved in PPIs during the CG simulations are shown as red sticks. (c) PPIs hotspots in the presence of ARG 10% w/w. (d) PPI hotspots without excipients. The catalytic domain of Reteplase is marked with an orange circle.

Figure 4

Representative RMSD vs. time plot of Reteplase during the FD simulations. Room temperature (RT), freezing (F), primary drying (1D), secondary drying (2D), and reconstitution (REC) denote the different steps of the FD process, which are separated by vertical lines. The reference state of the RMSD calculation is the starting structure of the simulations.

Figure 5

The plot of the protein radius of gyration. Three systems with different excipients at different concentrations are plotted: (a) Reteplase with ARG, (b) Reteplase with TXA, and (c) Reteplase with SUC. Each time course is obtained from the average of two independent simulations. Room temperature (RT), freezing (F), primary drying (1D), secondary drying (2D), and reconstitution (REC) denote the different steps of the FD process, which are separated by vertical lines.

Figure 6

The time course of the ratio between non-polar and total solvent accessible surface area (SASA). Three systems with different excipients at different concentrations are plotted: (a) Reteplase with ARG, (b) Reteplase with TXA, and (c) Reteplase with SUC. Each time course is obtained from the average of two independent simulations. Room temperature (RT), freezing (F), primary drying (1D), secondary drying (2D), and reconstitution (REC) denote the different steps of the FD process, which are separated by vertical lines.

Figure 7

The domain separation between the kringle-2 and catalytic domains after the reconstitution. The representative structures are obtained from the final frame of the simulations. The structures are aligned on the catalytic domain. The following structures are shown: Reteplase with: (a) no excipient, (b) ARG 10% w/w, (c) TXA 10% w/w, and (d) SUC 10% w/w. The connecting loop between the two domains is highlighted with an orange circle. The excipient located close to the connecting loop is shown as red sticks.

Figure 8

The difference scores between aggregation-prone residues (APR). The scores are calculated using A3D [56] and are based on taking the average of scores determined from the final frame of the two replicates. A negative score indicates that the residue becomes more aggregation prone. (a) The outputs from the RT simulations are compared to the secondary drying (2D) and reconstitution simulations (REC) without excipients. The local stabilizing effect of excipients (10% w/w) is monitored during: (b) 2D simulations and (c) REC simulations. The APRs that become more pronounced during the drying simulations are highlighted with colored letters (A–D). (d) The same APRs are highlighted in the Reteplase structure. (e,f) Effect of local excipient interactions. Lys40 in the kringle-2 region can interact with the carboxylic group of excipients. (e) Interaction between Lys40 and an ARG molecule is shown as an example. (f) Lys40 interacts with Asp64 and Asp66 in the presence of SUC. The snapshot is taken from the final frame of the reconstitution simulation with 10% w/w excipients. The local aggregation-prone region (Met35-Val41) is colored red.

Figure 9

Effect of ARG and TXA in low and high concentrations on the thermal stability of Reteplase. (a) Thermal unfolding profiles of Reteplase with and without excipients measured with nanoDSF. (b) Aggregation during heating measured with the backscattering detector in nanoDSF. All curves are mean of triplicates. Due to aggregation after sample preparation, the protein concentration in the conditions marked with (*) was lower.

Figure 10

Freeze-thaw and lyophilization stress-induced aggregation of Reteplase. (a) SEC chromatograms of Reteplase without excipient before (t0) and after FT and LYO. (b) Monomer recovery (%) of Reteplase detected by SEC. Time point t0 was set to 100% for each condition.

Figure 11

Effect of FT- and LYO-stress on Reteplase in the absence and presence of ARG and TXA measured by near-UV CD. (a) Spectrum of Reteplase with 10 mM ARG, (b) with 100 mM ARG, (c) with 10 mM TXA, (d) with 100 mM TXA, and (e) without an excipient.