Diselenide crosslinks for enhanced and simplified oxidative protein folding

Protein oxidative folding couples conformational folding with formation and reshuffling of disulfide bonds, often proceeding slowly and via heterogeneous intermediates that can trap scrambled isomers and reduce yield. Hirudin, a 65-residue, disulfide-rich thrombin inhibitor, exemplifies an extreme of complex, heterogeneous oxidative folding compared with BPTI’s native-like pathway. The study investigates whether replacing selected disulfide bonds with diselenide bonds (via cysteine to selenocysteine substitutions) can simplify hirudin’s folding landscape, enhance folding rates and yields, and preserve native structure and biological function. The central hypothesis is that preinstalled diselenide crosslinks at native or strategically chosen non-native positions would bias folding toward productive pathways, reducing intermediate heterogeneity and accelerating attainment of the native state.

Prior work delineates divergent oxidative folding mechanisms of disulfide-rich proteins, with BPTI representing native-like intermediate accumulation and hirudin following a heterogeneous, trial-and-error pathway with numerous scrambled species. Selenium chemistry has been leveraged to catalyze thiol/disulfide exchange and to direct protein folding using small-molecule diselenides and selenocysteine substitutions. In BPTI, diselenide substitutions (native and non-native) facilitated folding and bypassed common kinetic traps. Related studies on apamin, conotoxins, and insulin showed case-dependent impacts of diselenide incorporation on folding efficiency, stability, and activity. These precedents motivated assessing diselenide substitutions in hirudin, which has a markedly different, more chaotic folding pathway than BPTI, to evaluate generality and potential advantages for disulfide-rich protein preparation.

• Chemical synthesis: WT hirudin (WT-Hir) and four seleno-hirudin (Se-Hir) analogues were prepared by native chemical ligation (NCL) from two segments ligated at Gln38–Cys39. C-terminal peptides Hir(39–65) bearing Cys or Sec at position 39 were synthesized by Fmoc-SPPS, purified and characterized by HPLC and ESI-MS. N-terminal peptides Hir(1–38)-thioester surrogates (hydrazide or N-acylurea/Nbz) were prepared by Fmoc-SPPS, incorporating selenocysteine at specific sites (6, 14, 16, 22, 28) as needed. Cys/Sec-NCL was performed in degassed buffer (100 mM phosphate, 6 M guanidine HCl, 50–200 mM MPAA, pH ~7.3). Se-Hir analogues were obtained with one preformed diselenide and four free thiols; WT-Hir was isolated fully reduced.
• Analogues: Three with native diselenide bridges—6–14 [Hir(C6U/C14U)], 16–28 [Hir(C16U/C28U)], 22–39 [Hir(C22U/C39U)]—and one with a non-native bridge—6–16 [Hir(C6U/C16U)].
• Oxidative folding assays: Conducted mainly under anaerobic conditions (100 mM Tris-HCl, 200 mM NaCl, 1 mM EDTA, pH 8.7) with 30 µM protein and 150 µM GSSG at room temperature. Aliquots were quenched with HCl and analyzed by RP-HPLC (Atlantis T3 column, UV 220 nm) to monitor reduced (R) and 1-SS, 2-SS, 3-SS ensembles and native (N) species. Selected aerobic, no-GSSG folding experiments compared WT-Hir and Hir(C16U/C28U).
• Structural studies: Co-crystallization of Se-Hir analogues with bovine thrombin to obtain complexes: Complex-1 (Hir(C16U/C28U)-thrombin), Complex-2 (Hir(C6U/C14U)-thrombin), Complex-3 (Hir(C22U/C39U)-thrombin). X-ray structures were solved and compared with WT-Hir–thrombin structure (PDB 1HRT). RMSD values and local interactions were analyzed.
• NMR: 2D 1H DQF-COSY spectra of free WT-Hir, Hir(C6U/C14U), and Hir(C6U/C16U) in aqueous solution (pH 4.5) to assess structural similarity in solution.
• Thrombin inhibition assays: Bovine thrombin activity monitored at 405 nm using Tos-Gly-Pro-Arg-pNA substrate at 37 °C in Tris buffer with salts and PEG. Inhibitor concentrations titrated to derive apparent Ki values by fitting the provided inhibition equation. Protein concentrations determined spectrophotometrically.

• Folding acceleration and yield improvement:
  • WT-Hir (anaerobic, +GSSG): folding time 420 min, native yield 88% by HPLC (isolated 61 ± 1%), Ki = 10.9 ± 4.9 pM; HPLC native r.t. 11.2 min.
  • Hir(C16U/C28U): folding time 240 min, yield 90% (56 ± 1%), Ki = 10.0 ± 3.7 pM; native r.t. 11.3 min. Under aerobic conditions without GSSG: 80% folded after 300 min vs WT-Hir 27% after 1320 min.
  • Hir(C22U/C39U): folding time 90 min, yield 94% (61 ± 1%), Ki = 12.5 ± 2.9 pM; native r.t. 11.3 min. A predominant 3-SS intermediate peak observed (~23 min HPLC r.t.).
  • Hir(C6U/C14U): folding time 90 min, yield 95% (66 ± 1%), Ki = 192.4 ± 21.9 pM; native r.t. 14.8 min. Notably, 1-SS intermediates were not detected, indicating bypass of early folding ensembles.
  • Hir(C6U/C16U) (non-native diselenide): folding time 240 min, yield 90% (46 ± 1%), Ki = 104.9 ± 15.0 pM; native r.t. 13.2 min. Followed a heterogeneous pathway (mainly 2-SS and 3-SS) but converted efficiently to native.
• Folding pathway simplification: All Se-Hir analogues exhibited fewer, more distinct intermediate peaks and faster progression to native state than WT-Hir, reducing heterogeneity. In Hir(C6U/C14U) and partially in Hir(C6U/C16U), 1-SS populations were largely absent.
• Structural conservation: X-ray structures of Se-Hir–thrombin complexes were highly similar to WT-Hir–thrombin (PDB 1HRT) with overall RMSDs of 1.620 Å (C16U/C28U), 1.476 Å (C6U/C14U), and 1.472 Å (C22U/C39U). Intramolecular hydrogen-bond networks, especially around the central 6–14 crosslink, were preserved. PDB IDs: 7A0D (Complex-1), 7A0E (Complex-2), 7A0F (Complex-3).
• Activity: Native diselenide substitutions at 16–28 and 22–39 retained thrombin inhibition comparable to WT. Substitutions involving positions 6 and/or 14 or 16 (6–14 and 6–16) showed reduced activity (10–20-fold higher Ki), attributed to subtle local structural perturbations near these sites.
• Chromatographic behavior: Se-Hir variants with substitutions at/near positions 6, 14, 16 eluted later (higher r.t.), likely reflecting altered hydrophobicity under acidic HPLC conditions.

Introducing diselenide crosslinks into hirudin biased oxidative folding toward productive routes, reducing ensemble heterogeneity and accelerating attainment of the native state. This directly addresses the challenge of slow, heterogeneous oxidative folding typical of hirudin by stabilizing key crosslinks and facilitating reshuffling. Native-position diselenides at 16–28 and 22–39 delivered faster folding with yields and inhibitory activity similar to WT-Hir, while preserving global structure, supporting the hypothesis that diselenide substitution can streamline folding without compromising function. Substitutions involving the central 6–14 crosslink (and non-native 6–16) also simplified early folding (largely eliminating 1-SS ensembles) but modestly impaired activity, consistent with local conformational adjustments near a pivotal hydrogen-bonding cluster. Structural analyses confirmed that overall folds in complex with thrombin remained essentially unchanged, with small, localized differences around the substituted crosslinks. Aerobic, oxidant-free folding further highlighted the catalytic advantage of Se substitution. Collectively, results generalize prior benefits observed in other proteins to a model with a notoriously heterogeneous pathway, underscoring the utility of disulfide-to-diselenide replacement for improving foldability of disulfide-rich proteins.

This work demonstrates that strategic diselenide substitutions in hirudin simplify oxidative folding pathways, accelerate folding, and increase yields while largely preserving native structure and, in most cases, biological activity. Native diselenides at 16–28 and 22–39 retained thrombin inhibition comparable to WT-Hir; substitutions involving the central 6–14 (and non-native 6–16) efficiently streamlined folding but slightly reduced activity due to local structural perturbations. Crystal structures of Se-Hir–thrombin complexes confirmed high structural conservation relative to WT. These findings, together with prior studies on BPTI, conotoxins, and insulin, support disulfide-to-diselenide replacement as a broadly useful strategy for design, preparation, and characterization of disulfide-rich proteins. Potential future directions include systematic mapping of substitution positions to optimize both folding efficiency and function across diverse disulfide-rich proteins, extending studies to free (unbound) states and in vivo contexts, and leveraging diselenide substitutions to rationally tune stability, specificity, and activity for therapeutic applications.

• Structural comparisons were performed on thrombin-bound complexes; bound conformations may not fully reflect free-solution structures, although 2D 1H-NMR COSY indicated similar solution structures for selected analogues.
• Some trapped scrambled intermediates persisted in Se-Hir folding (e.g., Hir(C16U/C28U)), indicating that not all non-productive species are eliminated by Se substitution.
• Substitutions at or near positions 6, 14, and 16, while simplifying folding, reduced inhibitory activity, highlighting position-dependent trade-offs between foldability and function.
• Free hirudin proteins were challenging to crystallize; structural inferences rely on inhibitor–enzyme complexes.