Diselenide-bond replacement of the external disulfide bond of insulin increases its oligomerization leading to sustained activity

Insulin is a double-chain peptide hormone stabilized by three disulfide bonds. Substitution of cysteine residues with selenocysteine to form diselenide bonds can alter folding kinetics and thermodynamic stability of proteins. The authors previously reported a bovine pancreatic insulin analog in which the external CysA7–CysB7 disulfide is replaced by a diselenide ([C7UA,C7UB] SeIns), showing increased resistance to insulin-degrading enzyme (IDE). A second, independently reported human insulin analog with an internal Se–Se ([C6UA,C11UA]) also showed enhanced stability. However, Weiss et al. questioned the IDE resistance and in vivo protraction of SeIns. The present study investigates whether SeIns indeed exhibits superior IDE resistance and sustained hypoglycemic action, and explores the molecular basis—testing the hypothesis that replacing the solvent-exposed external S–S with Se–Se promotes oligomerization, thereby protecting insulin from IDE and prolonging activity.

Past synthetic advances include stepwise disulfide formation using orthogonally protected chains, single-chain proinsulin mimics, and native chain assembly (NCA). Selenium substitution strategies can enhance folding and stability due to lower Se–Se reduction potential. The first seleno-insulin ([C7UA,C7UB], external Se–Se) was reported to resist IDE, potentially via monomer stabilization at the B-chain N-terminus. A second analog ([C6UA,C11UA], internal Se–Se) improved thermodynamic and kinetic stability against proteolysis and reductive unfolding. Weiss et al. re-evaluated the external Se–Se analog and did not observe significant IDE resistance or sustained action, highlighting discrepancies possibly due to differing assay conditions and synthesis routes. Human insulin oligomerizes above ~10 µM, and IDE accepts only monomeric insulin into its catalytic chamber, suggesting oligomerization may underlie protease resistance.

Synthesis via native chain assembly (NCA): Synthetic SeIns A- and B-chains were prepared with protective groups: SeInsA[1SeS,2SPys] and SeInsB either [SeS] or [SPys,SePys]. Chains (200 µM each) were combined in 25 mM sodium bicarbonate buffer (pH 10.0) with urea (0.36–0.48 M), EDTA (1 mM), DTTred (4.8 mM; added from 34.3 mM stock), 10% ethylene glycol, with or without protein disulfide isomerase (PDI, 4.0 µM). Reactions were incubated at low temperature (−10 °C nominal; 4–10 °C instrument) for 1–6 days. Progress was monitored by RP-HPLC (Tosoh TSKgel ODS-100V) after quenching with AEMTS, and products characterized by HPLC retention time, MALDI-TOF-MS, and amino acid analysis. Yields were calculated from HPLC peak areas. IDE degradation assays: Human IDE was incubated with insulin substrates under varying S/E conditions: 5.0 µM insulin with 50 nM IDE (S/E=100:1) and 20 µM insulin with 100 nM IDE (S/E=200:1) in 0.1 M Tris-HCl, pH 8.0, at 30 °C. Single-component assays (BPIns or SeIns) and competition assays using mixtures (BPIns:SeIns = 1:1, 1:4, 4:1; total 5 µM) were performed. Reactions were quenched with 1 M HCl and analyzed by RP-HPLC (COSMOSIL 5C18-AR-II or Tosoh TSKgel ODS-100V) with UV detection at 220 nm. Remaining substrate versus time was fit to a single-exponential decay to obtain apparent first-order rate constants (kapp) and half-lives. Analytical ultracentrifugation (AUC): Sedimentation velocity experiments (60,000 rpm, 20 °C, absorbance at 230 nm) in 0.1 M Tris-HCl pH 8.0 examined BPIns, SeIns, and 1:1 mixtures at 5, 10, and 20 µM. Data were analyzed with SEDFIT to obtain c(S) distributions; solvent parameters from SEDNTERP. Circular dichroism (CD): Far-UV CD spectra were recorded (0.1 cm pathlength). Thermal unfolding was assessed from 5 to 110 °C in 20 mM sodium phosphate pH 7.5 containing 1 M Gdn-HCl to facilitate unfolding. Ellipticity at 222 nm versus temperature was fit under a reversible two-state assumption to estimate apparent Tm, ΔH, and ΔCp, acknowledging observed irreversibility on cooling. In vivo hypoglycemic studies: Normal and streptozotocin-induced diabetic rats received subcutaneous injections of insulin formulations dissolved in saline: rapid-acting lispro (Humalog), synthetic human insulin (HIns), synthetic bovine pancreatic insulin (BPIns), or SeIns. Doses were 15 µg/300 g rat (10 U/mL) and 150 µg/300 g rat (100 U/mL). Blood glucose was monitored over time using an i-STAT 1 analyzer (cartridge 6+) from tail vein blood. Survival and short-term toxicity were assessed (Supplementary Table 1). Additional assays assessed reduction by glutathione (GSH) and degradation in human serum at 37 °C with RP-HPLC quantification.

• Optimized NCA markedly improved SeIns two-chain folding yields: from 27% (prior) to 52–56% under optimized conditions, and up to 70–72% using SeInsB protected with 2-pyridylsulfanyl groups, comparable to in vitro proinsulin folding yields.
• IDE resistance: At [S]=5.0 µM, [E]=50 nM, SeIns degraded significantly slower than BPIns (T1/2≈7.6 h for SeIns vs 2.4 h for BPIns). Under Weiss et al. conditions ([S]=20 µM, [E]=100 nM), kapp for SeIns was 0.042 h−1 (T1/2≈17.2 h) versus BPIns 0.140 h−1 (T1/2≈5.2 h), confirming superior resistance across conditions.
• Mixed-substrate assays showed SeIns presence slowed BPIns degradation; in 1:1 mixtures at total 5 µM, kappBPIns≈0.097 h−1 and kappSeIns≈0.095 h−1, with total decay similar to SeIns alone. Even at BPIns:SeIns=4:1, BPIns degradation was notably decelerated, indicating hetero-oligomer formation that limits monomer availability to IDE.
• AUC demonstrated enhanced oligomerization by SeIns: at 5 µM, SeIns already showed dimer (≈2 S) whereas BPIns was predominantly monomer (<1 S). At 20 µM, SeIns exhibited higher dimer populations and detectable hexamer (~4 S) even without Zn2+. Mixtures displayed distributions inconsistent with simple sums, supporting hetero-oligomer formation.
• CD and thermal stability: Both insulins are predominantly α-helical; SeIns exhibited a slightly lower estimated α-helix content by BeStSel (32.7% vs 36.5% for BPIns), consistent with differing oligomerization states. Thermal denaturation (with 1 M Gdn-HCl) yielded higher apparent stability for SeIns (Tm 72.6 ± 0.2 °C; ΔH 24.5 ± 6.1 kcal/mol) than BPIns (Tm 58.4 ± 1.3 °C; ΔH 19.6 ± 1.2 kcal/mol).
• In vivo pharmacology: At 15 µg/300 g, SeIns behaved similarly to BPIns and HIns in normal and diabetic rats. At 150 µg/300 g, SeIns displayed a clear long-acting hypoglycemic effect in both normal and diabetic rats, with low short-term toxicity comparable to HIns.
• Stability in biological milieu: Reduction by GSH was similar for SeIns and BPIns, indicating that thiol reduction resistance is not the primary driver of protraction. In human serum, SeIns degraded slightly more slowly (kobs≈0.015 h−1; T1/2≈46 h) than BPIns (kobs≈0.020 h−1; T1/2≈35 h), a modest difference insufficient alone to explain in vivo protraction.
• Mechanistic interpretation: A preequilibrium between oligomers and monomers, with SeIns exhibiting stronger oligomerization and slightly greater monomer stability, reduces monomer availability to IDE and slows clearance, accounting for sustained activity.

The study addresses whether substituting the solvent-exposed external disulfide (CysA7–CysB7) with a diselenide confers genuine IDE resistance and sustained in vivo activity. The consistent finding of slower IDE-catalyzed degradation for SeIns across substrate/enzyme concentrations, together with competition experiments, supports a mechanism where enhanced oligomerization—both homo- and hetero- with wild-type insulin—reduces monomer availability to IDE. AUC directly visualizes increased oligomer formation by SeIns and hetero-oligomers in mixtures, aligning with the preequilibrium model. Increased apparent thermal stability further suggests slightly more stable monomeric or oligomeric conformations. In vivo, SeIns shows protraction at higher dosing, likely due to slower dissociation of subcutaneous oligomers and reduced hepatic IDE-mediated clearance, rather than increased resistance to thiol-mediated reduction or markedly improved serum stability. These results reconcile prior conflicting reports by emphasizing assay condition dependence and the dominant role of oligomerization dynamics. The findings are relevant for insulin formulation design, offering a chemical approach—Se substitution at a solvent-exposed bond—to modulate association behavior and protease susceptibility without impairing bioactivity.

Applying optimized native chain assembly enabled high-yield synthesis (up to 72%) of a seleno-insulin analog bearing an external diselenide bridge. SeIns exhibits superior resistance to IDE, primarily due to a strong propensity to form soluble higher-order oligomers and hetero-oligomers with wild-type insulin, and shows higher apparent thermal stability. In rats, SeIns demonstrates sustained hypoglycemic action at higher doses. Replacing the solvent-exposed S–S with Se–Se emerges as a viable strategy to enhance insulin persistence by tuning oligomerization and protease resistance. Future work should include detailed pharmacokinetics, long-term toxicity assessment, and exploration of co-formulations or additional SeIns variants to optimize protraction and safety.

• Publication reports no detailed pharmacokinetic parameters; the mechanistic attribution to slowed subcutaneous dissociation and IDE resistance requires PK/PD studies.
• Thermal unfolding analyses assumed reversibility despite observed irreversibility upon cooling, so thermodynamic parameters are apparent.
• AUC measurements were limited to concentrations ≥5 µM; behavior at lower, physiological concentrations was inferred rather than directly observed.
• In vivo efficacy was demonstrated in rat models with sustained action evident only at higher dose (150 µg/300 g); translation to human dosing and long-term safety remain untested.
• Serum stability differences were modest, implying other unmeasured factors may contribute to protraction.
• Long-term toxicity and immunogenicity of selenium substitution were not evaluated.