Search

Daily injections of insulin provide lifesaving benefits to millions of diabetics. But currently available prandial insulins are suboptimal: The onset of action is delayed by slow dissociation of the insulin hexamer in the subcutaneous space, and insulin forms amyloid fibrils upon storage in solution. Here we show, through the use of noncanonical amino acid mutagenesis, that replacement of the proline residue at position 28 of the insulin B-chain (ProB28) by (4S)-hydroxyproline (Hzp) yields an active form of insulin that dissociates more rapidly, and fibrillates more slowly, than the wild-type protein. Crystal structures of dimeric and hexameric insulin preparations suggest that a hydrogen bond between the hydroxyl group of Hzp and a backbone amide carbonyl positioned across the dimer interface may be responsible for the altered behavior. The effects of hydroxylation are stereospecific; replacement of ProB28 by (4R)-hydroxyproline (Hyp) causes little change in the rates of fibrillation and hexamer disassociation. These results demonstrate a new approach that fuses the concepts of medicinal chemistry and protein design, and paves the way to further engineering of insulin and other therapeutic proteins.

Insulin remains a challenging synthetic target due in large part to its two-chain, disulfide-constrained structure. Biomimetic single chain precursors inspired by proinsulin that utilize short peptides to join the A and B chains can dramatically enhance folding efficiency. Systematic chemical analysis of insulin precursors using an optimized synthetic protocol identified a 49 amino acid peptide named DesDi, which folds with high efficiency by virtue of an optimized structure and could be proteolytically converted to bioactive two-chain insulin. In subsequent applications, we observed that the folding of the DesDi precursor was highly tolerant to amino acid substitution at various insulin residues. The versatility of DesDi as a synthetic insulin precursor was demonstrated through the preparation of several alanine mutants (A10, A16, A18, B12, B15), as well as ValA16, an analog that was unattainable in prior reports. In vitro bioanalysis highlighted the importance of the native, hydrophobic residues at A16 and B15 as part of the core structure of the hormone and revealed the significance of the A18 residue to receptor selectivity. We propose that the DesDi precursor is a versatile synthetic intermediate for the preparation of diverse insulin analogs. It should enable a more comprehensive analysis of function to insulin structure than might not be otherwise possible through conventional approaches.

Systemic amyloidoses, an important class of protein misfolding diseases, are often due to fibrillation of disulfide-cross-linked globular proteins otherwise unrelated in sequence or structure. Although cross-beta assembly is regarded as a universal property of polypeptides, it is not understood how such amyloids accommodate diverse disulfide connectivities. Does amyloidogenicity depend on protein topology? A model is provided by insulin, a two-chain protein containing three disulfide bridges. The importance of chain topology is demonstrated by mini-proinsulin (MP), a single-chain analogue in which the C-terminus of the B chain (residue B30) is tethered to the N-terminus of the A chain (A1). The B30-A1 tether impedes the fiber-specific alpha --> beta transition, leading to slow formation of a structurally nonuniform amorphous precipitate. Conversely, fibrillation is robust to interchange of disulfide bridges. Whereas native insulin exhibits pairings [A6-A11, A7-B7, and A20-B19], metastable isomers with alternative pairings [A6-B7, A7-A11, A20-B19] or [A6-A7, A11-B7, A20-B1] readily undergo fibrillation with essentially identical alpha --> beta transitions. Respective pairing schemes are in each case retained. Isomeric fibrils and the amorphous MP precipitate are each able to seed the fibrillation of wild-type insulin, suggesting a structural correspondence between respective nuclei or modes of assembly. Together, our results demonstrate that effects of polypeptide topology on amyloidogenicity depend on structural context. Although the native structures and stabilities of single-chain insulin analogues are similar to those of wild-type insulin, the interchain tether constrains the extent of conformational distortion at elevated temperature, retards initial non-native aggregation, and is apparently incompatible with the mature structure of an insulin protofilament. We speculate that the general danger of fibrillation has imposed a constraint in protein evolution, selecting for topologies unfavorable to amyloid formation.

Although insulin and insulin-like growth factor 1 (IGF-1) share homologous sequence, similar tertiary structure, weakly overlapped biological activity, and a common ancestor, the two highly homologous sequences encode different folding behavior: insulin folds into one unique stable tertiary structure while IGF-1 folds into two disulfide isomers with similar thermodynamic stability. To further elucidate the molecular mechanism of their different folding behavior, we prepared two single-chain hybrids of insulin and IGF-1, Ins(A)/IGF-1(B) and Ins(B)/IGF-1(A), as well as a mini-IGF-1 by means of protein engineering and studied their structure as well as folding behavior. Both mini-IGF-1 and Ins(A)/IGF-1(B) fold into two thermodynamically stable disulfide isomers in vivo and in vitro just like that of IGF-1, while Ins(B)/IGF-1(A) folds into one unique thermodynamically stable tertiary structure in vivo and in vitro just like that of insulin. So we deduce that the different folding behavior of insulin and IGF-1 is mainly controlled by their B-chain/domain. By V8 endoproteinase digestion and circular dichroism analysis, as well as insulin receptor binding assay, we deduce that Ins(B)/IGF-1(A), isomer 2 of mini-IGF-1, and isomer 2 of Ins(A)/IGF-1(B) adopt native IGF-1/insulin-like three-dimensional structure with native disulfides, while isomer 1 of mini-IGF-1 and isomer 1 of Ins(A)/IGF-1(B) adopt the swap IGF-1-like three-dimensional structure with swap disulfides.

The landscape paradigm of protein folding can enable preferred pathways on a funnel-like energy surface. Hierarchical preferences may be manifest as a nonrandom pathway of disulfide pairing. Stepwise stabilization of structural subdomains among on-pathway intermediates is proposed to underlie the disulfide pathway of proinsulin and related molecules. Here, effects of pairwise serine substitution of insulin's exposed interchain disulfide bridge (Cys(A7)-Cys(B7)) are characterized as a model of a late intermediate. Untethering cystine A7-B7 in an engineered monomer causes significantly more marked decreases in the thermodynamic stability and extent of folding than occur on pairwise substitution of internal cystine A6-A11 [Weiss, M. A., Hua, Q. X., Jia, W., Chu, Y. C., Wang, R. Y., and Katsoyannis, P. G. (2000) Biochemistry 39, 15429-15440]. Although substantially disordered and without significant biological activity, the untethered analogue contains a molten subdomain comprising cystine A20-B19 and a native-like cluster of hydrophobic side chains. Remarkably, A and B chains make unequal contributions to this folded moiety; the B chain retains native-like supersecondary structure, whereas the A chain is largely disordered. These observations suggest that the B subdomain provides a template to guide folding of the A chain. Stepwise organization of insulin-like molecules supports a hierarchic view of protein folding.

Although the structure of insulin has been well studied, the formation pathway of the three disulfide bridges during the refolding of insulin precursor is ambiguous. Here, we reported the in vitro disulfide-forming pathway of a recombinant porcine insulin precursor (PIP). In redox buffer containing L-arginine, the yield of native PIP from fully reduced/denatured PIP can reach 85%. The refolding process was quenched at different time points, and three distinct intermediates, including one with one disulfide linkage and two with two disulfide bridges, have been captured and characterized. An intra-A disulfide bridge was found in the former but not in the latter. The two intermediates with two disulfide bridges contain the common A20-B19 disulfide linkage and another inter-AB one. Based on the time-dependent formation and distribution of disulfide pairs in the trapped intermediates, two different forming pathways of disulfide bonds in the refolding process of PIP in vitro have been proposed. The first one involves the rapid formation of the intra-A disulfide bond, followed by the slower formation of one of the inter-AB disulfide bonds and then the pairing of the remaining cysteines to complete the refolding of PIP. The second pathway begins first with the formation of the A20-B19 disulfide bridge, followed immediately by another inter-AB one, possibly nonnative. The nonnative two-disulfide intermediates may then slowly rearrange between CysA6, CysA7, CysA11, and CysB7, until the native disulfide bond A6-A11 or A7-B7 is formed to complete the refolding of PIP. The proposed refolding behavior of PIP is compared with that of IGF-I and discussed.

The intracellular conversion of proinsulin to insulin occurs via cleavage at the two dibasic sites: Arg31-Arg32, B chain-C-peptide (BC) junction; and Lys64-Arg65, A chain-C-peptide (CA) junction, catalyzed by the subtilisin-like prohormone convertases SPC3 (PC1/PC3) and SPC2 (PC2), respectively. In this report we propose a possible conformational variant of proinsulin that would facilitate the formation of enzyme-substrate complexes at the BC and AC junctions of proinsulin with the substrate binding groove of the two closely related convertases. Productive convertase interaction requires extended peptide conformations in both the CA junction (residues 62-67, LQKRGI) and the BC junction (residues 29-34, KTRREA) and leads to significant perturbations in the normally alpha-helical N-terminal region of the A chain and the extended C-terminal region of the B chain of the insulin moiety of proinsulin. In this model of the reactive conformation of human proinsulin, both processing sites assume positions that are relatively far apart. The C-peptide was then modeled in an unobtrusive conformation relative to the convertases and the remainder of the substrate, forming an extended loop of length approximately 40 A with a short alpha-helical segment rather than a random coil. A model of the stereochemical transformations that occur during the processing of proinsulin by SPC2 is presented.

The proinsulin-insulin system provides a general model for the proteolytic processing of polypeptide hormones. Two proinsulin-specific endopeptidases have been defined, a type I activity that cleaves the B-chain/C-peptide junction (Arg31-Arg32) and a type II activity that cleaves the C-peptide/A-chain junction (Lys64-Arg65). These endopeptidases are specific for their respective dibasic target sites; not all such dibasic sites are cleaved, however, and studies of mutant proinsulins have demonstrated that additional sequence or structural features are involved in determining substrate specificity. To define structural elements required for endopeptidase recognition, we have undertaken comparative 1H NMR and photochemical dynamic nuclear polarization (photo-CIDNP) studies of human proinsulin, insulin, and split proinsulin analogues as models of prohormone processing intermediates. The overall conformation of proinsulin is observed to be similar to that of insulin, and the connecting peptide is largely unstructured. In the 1H NMR spectrum of proinsulin significant variation is observed in the line widths of insulin-specific amide resonances, reflecting exchange among conformational substates; similar exchange is observed in insulin and is not damped by the connecting peptide. The aromatic 1H NMR resonances of proinsulin are assigned by analogy to the spectrum of insulin, and assignments are verified by chemical modification. Unexpectedly, nonlocal perturbations are observed in the insulin moiety of proinsulin, as monitored by the resonances of internal aromatic groups. Remarkably, these perturbations are reverted by site-specific cleavage of the connecting peptide at the CA junction but not the BC junction. These results suggest that a stable local structure is formed at the CA junction, which influences insulin-specific packing interactions. We propose that this structure (designated the "CA knuckle") provides a recognition element for type II proinsulin endopeptidase.