Diselenide crosslinks for enhanced and simplified oxidative protein folding

The *in vitro* oxidative folding of proteins, a process involving both conformational folding and disulfide bond formation, has been extensively studied. However, many proteins, especially those rich in disulfide bonds, exhibit slow and inefficient folding *in vitro*, posing challenges for the production of therapeutic proteins. This process often involves the formation of numerous heterogeneous intermediates, leading to reduced yield and slower kinetics. Two extreme models representing this diversity are bovine pancreatic trypsin inhibitor (BPTI), which folds via a relatively simple pathway with few distinct intermediates, and hirudin, a potent thrombin inhibitor characterized by a highly heterogeneous “trial-and-error” folding pathway involving many scrambled isomers. Hirudin’s complex folding pathway, involving numerous disulfide-bonded intermediates, makes it a challenging model system for studying oxidative protein folding but also makes it an excellent model for exploring strategies to improve this process. This study uses hirudin as a model system to test the hypothesis that incorporating diselenide bridges, which are known to catalyze disulfide exchange reactions, would improve the efficiency of hirudin’s oxidative folding by reducing the number of unproductive folding intermediates. This rationale is supported by prior work showing that selenocysteine substitutions in other proteins, such as BPTI, improved folding efficiency. The researchers expected that minimizing random disulfide pairing, inherent in hirudin's native folding, would favor productive folding pathways over unproductive ones, resulting in faster and more efficient folding.

The literature review covers existing knowledge on oxidative protein folding, highlighting the contrasting folding mechanisms of BPTI and hirudin. It mentions the use of selenium-containing molecules and selenocysteine substitutions as strategies to improve oxidative folding in other proteins, particularly in the context of improving efficiency. Previous research on BPTI showed that seleno-analogues folded correctly and bypassed common intermediates, providing the basis for investigating diselenide substitutions in hirudin. The review also includes details about the complexities of hirudin's folding mechanism and the negative impacts of accumulating scrambled isomers. Studies on other proteins, including apamin and BPTI, where non-native diselenide bridges were incorporated, showed mixed results. While apamin folding was impeded by the non-native diselenide, BPTI analogues folded correctly. This underscores the need to investigate the effects of diselenide positions within the protein.

The study employed chemical synthesis to create wild-type hirudin (WT-Hir) and four seleno-hirudin (Se-Hir) analogues. The analogues were designed with diselenide bonds replacing native disulfide bonds at positions 6-14, 16-28, and 22-39, as well as one analogue with a non-native diselenide bond at position 6-16. Native chemical ligation (NCL) was the primary method for synthesizing these proteins. The N- and C-terminal peptides were synthesized separately using Fmoc-SPPS, and then ligated together. Oxidative folding experiments were conducted under both anaerobic conditions, using oxidized glutathione (GSSG) as an oxidant, and aerobic conditions without GSSG. HPLC was used to monitor the folding process and quantify the amounts of different intermediates and the native state at various time points. The kinetics of thrombin inhibition by the different analogues was measured, providing a functional assessment. Finally, X-ray crystallography was used to analyze the three-dimensional structures of the Se-Hir analogues complexed with thrombin, to determine the effect of the diselenide substitutions on protein structure. 2D ¹H-NMR COSY analysis was performed on free WT-Hir and two Se-Hir analogues to compare their solution structure.

The introduction of diselenide crosslinks at native positions generally enhanced the folding of hirudin, leading to increased folding rate and yield, alongside a reduction in the heterogeneity of folding intermediates. For example, the Hir(C16U/C28U) analogue folded five times faster than WT-Hir under anaerobic conditions with comparable yield. Under aerobic conditions without GSSG, the folding yield of Hir(C16U/C28U) was 80% after 5 hours compared to WT-Hir’s 27% after 22 hours. The most significant improvement in folding was shown by Hir(C22U/C39U), reaching 66% of the native state within 5 min under anaerobic conditions. Interestingly, for analogues Hir(C6U/C14U) and Hir(C6U/C16U), the 1-SS intermediates were largely absent, suggesting that early unproductive folding steps were bypassed. While most of the Se-Hir analogues maintained the overall 3D structure and thrombin inhibitory activity similar to WT-Hir, Hir(C6U/C14U) and Hir(C6U/C16U) exhibited a 20-fold and 10-fold decrease in inhibitory activity, respectively, possibly due to local structural changes around the substituted residues.

The results demonstrate that diselenide substitutions significantly enhance the efficiency of hirudin’s oxidative folding. The improvement arises primarily from a reduction in the heterogeneity of intermediates, leading to faster and more productive folding pathways. The success of the approach using non-native diselenide bonds (6-16) further underscores its versatility. The observed decrease in inhibitory activity for certain analogues, despite their efficient folding, emphasizes the importance of precise placement of diselenide bridges to fully retain function. The findings extend previous research demonstrating the beneficial effects of diselenide substitutions in other proteins, suggesting a general strategy for improving protein folding in basic and applied research. Although similar variations in chromatography were reported by Alewood et al. of the seleno-analogues of conotoxins, the differences observed in the retention time, combined with the lower activity against thrombin, raised concerns as to whether the Hir(C6U/C14U) analogue had correctly folded into the native state.

This study demonstrates that replacing disulfide bonds with diselenide bonds in hirudin significantly improves its *in vitro* oxidative folding, leading to faster kinetics and higher yields. While the majority of the analogues retained their biological activity, the specific location of the diselenide substitutions impacted both folding efficiency and functional activity, highlighting the importance of site selection. This approach has broad implications for protein engineering, offering a potential tool for the production of disulfide-rich therapeutic proteins. Future research could focus on exploring different diselenide bridge positions within hirudin and other complex proteins to optimize folding and functional outcomes.

The study primarily focuses on *in vitro* folding, and the results may not directly translate to the *in vivo* environment where chaperones and other cellular factors influence protein folding. The sample size for some of the analyses was limited and additional replicates may strengthen certain conclusions. The analysis focuses on the complex formed between hirudin and thrombin. Thus, the observation does not completely eliminate the possibility that the free forms of Hir(C6U/C14U) and Hir(C6U/C16U) may adopt a different structure outside the complex. However, 2D ¹H-NMR COSY analysis supported the conclusions regarding solution structures.