Articles pubs.acs.org/acschemicalbiology
Discovery of High Potency, Single-Chain Insulin Analogs with a Shortened B‑Chain and Nonpeptide Linker Zachary P. Kaur,† Alexander R. Ochman,‡ John P. Mayer,† Vasily M. Gelfanov,† and Richard D. DiMarchi*,† †
Department of Chemistry, Indiana University, Bloomington, Indiana 47405, United States Melior Discovery Inc., Exton, Pennsylvania 19341, United States
‡
S Supporting Information *
ABSTRACT: A series of novel, single chain insulin analogs containing polyethylene glycol based connecting segments were synthesized by native chemical ligation and tested for biological activity. While the full length single chain insulin analogs exhibited low potency, deletion of amino acids B26− B30 unexpectedly generated markedly higher activity. This observation is unprecedented in all previous studies of single chain insulin analogs and is consistent with the presumption that in the native hormone this sequence must translocate to achieve high potency insulin receptor interaction. Optimization of the sequence yielded an insulin analog with potency and selectivity comparable to that of native insulin. These results establish a basis for discovery of novel higher potency, single chain insulin analogs of shortened length.
I
function.6−13 The synthesis of single chain insulin analogs represents an opportunity to increase yields through intramolecular formation of the disulfide bonds in analogy to proinsulin.14 While these “mini-proinsulins” may subsequently require enzymatic or chemical conversion to two-chain analogs, there is precedent to suggest that single chain insulin analogs may potentially be of clinical utility.15 The initial single chain insulin analogs were prepared semisynthetically using aliphatic dicarboxylic acid-based linkers of 2−12 atoms that covalently attached the N-terminus of the A-chain to a site proximal to the C-terminus of the B-chain. The insulin analogs showed a range of activities of 2−40% when compared to native insulin and varied as a function of linkerlength.16 The direct linear linkage of LysB29 to GlyA1 produced an inactive analog that further highlighted the importance of spacer length and conformational flexibility in the B-chain C-terminus.17 Biosynthetic insulin analogs with various “mini” C-peptides have been produced in high yields, but most required excision of the C-peptide for full bioactivity. The report of Sohma et al. represented a leap forward in the chemical synthesis of an insulin-like peptide by employing native chemical ligation of the three peptide fragments.18 In that same year, a single chain insulin analog with a C-peptide linker sequence of GGGPGRR and four mutations in the A- and B-chain (HisA8, AspB10, AspB28, and ProB29) was reported to be of high potency relative to native insulin.19 These substitutions are Insulin-like Growth
nsulin is a 51 amino acid hormone containing two peptide chains connected by two disulfide bonds. The A-chain is 21 amino acids in length and the B-chain is 30 amino acids. The B-chain cysteines at positions 7 and 19 form disulfide bonds to the A-chain cysteines at positions 7 and 20, respectively, while the A-chain also contains an intramolecular disulfide bond between cysteines at 6 and 11 (Figure 1A). Proinsulin contains a 35 amino acid C-peptide that connects the C-terminus of B-chain and the N-terminus of A-chain. This peptide is proteolytically cleaved to yield the two-chain form of native insulin. The C-peptide assists in proper disulfide-pairing and also renders the molecule approximately 50-fold lower in potency at the insulin receptor.1 Native insulin is stored as an insoluble granule in the pancreas as a Zn2+stabilized hexamer that dissociates to monomers to bind to its receptor.1 Insulin replacement therapy is of life-saving importance to millions of diabetics and commercial biosynthesis has enabled the manufacture of native insulin as well as analogs with improved pharmacokinetic properties.2 While conventional rDNA-based biosynthesis is restricted to the 20 coded amino acids, chemical methods can access a broader diversity of synthetic building blocks and provide a route to insulin analogs with enhanced performance. Total chemical synthesis of insulin has presented a significant challenge since the first reports of its preparation nearly 50 years ago.3−5 The formation of the correct disulfides in the absence of a native connecting-peptide represents the primary obstacle to obtaining a high yield synthesis. Nonetheless, a variety of low-yield methods were used to make a large number of two chain analogs, which have collectively shaped our current understanding of the relationship of insulin structure to biological © 2013 American Chemical Society
Received: April 16, 2013 Accepted: June 3, 2013 Published: June 3, 2013 1822
dx.doi.org/10.1021/cb4002624 | ACS Chem. Biol. 2013, 8, 1822−1829
ACS Chemical Biology
Articles
Figure 1. (A) Representation of human insulin illustrating the two peptide chains connected by disulfide bonds. Conserved residues between insulin and IGF-1 are shaded in gray. (B) Native chemical ligation scheme for synthesizing analogs.
The rationale for employing the IGF-1 sequence was intended to enhance the quality of the synthesis and improve the biophysical character relative to native human insulin. IGF-1 is an endogenous protein with approximately 50% sequence homology to insulin (Figure 1A) that circulates in plasma at relatively high concentrations while bound to a carrier protein.21 It contains notable changes in the A-chain, which increase its relative hydrophilicity and solubility in physiological buffers. Sequence variation from insulin, particularly at residues B10, 28, and 29, collectively reduces selfassociation and serves to maintain IGF-1 in a monomeric state.20,22 The mutations within the midregion of the insulin B-chain of Tyr-Leu to Gln-Phe in IGF-1 effectively diminish activity at the insulin receptor, and its reversal provides high potency analogs.23,24 Figure 2A shows the primary sequences of all insulin analogs analyzed in this study. Five primary modifications included in all IGF-1 A- and B-chain sequences were introduced to optimize chemical stability and biological activity. Alanine scanning mutagenesis studies suggested that AsnA21 is of higher insulin receptor activity than the corresponding alanine of native IGF-1.25 Asparagine at A18 eliminated the methionine that is of minor importance to IGF-1 receptor binding and removed a potential oxidation site. Most importantly, the Tyr16 and Leu17 modifications noted above were included along with histidine at position B5 in order to enhance insulin receptor recognition.23,26 Additional modifications specific to certain analogs are discussed in later sections.
Factor 1 (IGF-1) related and were inspired by enhanced monomeric biophysical character.20 This investigation explores the application of nonpeptidyl linkers to the synthesis and biological study of single chain insulin analogs. Small molecular weight polyethylene glycol (PEG) linkers of defined length were assembled as part of a single chain insulin to explore the optimal distance for achieving biological activity, devoid of any secondary structures commonly associated with peptidyl linkers. The results were inconsistent with prior observations using peptidyl linkers, as each of the PEG-linked single chain insulins with full B-chain length demonstrated low insulin potency. Shortening the B-chain to what is commonly regarded as the minimal C-terminal length surprisingly revealed full biological potency that was responsive to linker length.
■
RESULTS AND DISCUSSION Single Chain Insulin Design Strategy. Chemical synthesis of single chain insulin is a relatively recent achievement with the vast majority of insulin analogs having been prepared either semisynthetically or through chain combination of A- and B-chains individually prepared by various procedures.2 The synthetic strategy utilized in this study is a one step, native chemical ligation of two peptide segments prepared by solid-phase chemical methods (Figure 1B). The initial single chain insulins were based upon the IGF-1 sequence with a subtle mutation in the B-chain to significantly increase activity at the insulin receptor. 1823
dx.doi.org/10.1021/cb4002624 | ACS Chem. Biol. 2013, 8, 1822−1829
ACS Chemical Biology
Articles
Figure 2. (A) Sequences of peptides analyzed by biological assays. Numbers correspond to compound numbers in Table 1. Residues conserved between IGF-1 and insulin (gray); IGF-1 (black); insulin (red); non-native to IGF-1 and insulin (blue). (−) Represents connection between chains. D-chain not shown on IGF-1 sequence. (B) PEG linkers used in the single chain analogs.
PEG was chosen as a linker due to its flexibility, lack of secondary structure, and availability in defined molecular lengths. PEG polymers (−CH2CH2O−)n of n = 4, 8, 12, and 16 (2 × 8) repeating units were used to explore the optimal linear distance to connect the A- and B-chains (Figure 2B). PEG4 (16 atom spacer) was the smallest linker used with a length which is intermediate between previously reported chemical linkers (8−22 atoms).16 PEG8, PEG12, and PEG16 linkers represent a broadened range of 28, 40, and 56 atoms, extending the molecular length beyond where bioactivity has been previously reported.15,19 The PEG16 is formed by connecting two PEG8 monomers and, as such, contains four extra atoms relative to a linear PEG16 (Figure 2B). The linkers were coupled as part of the solid-phase assembly of peptide fragment 1 connecting GlyA1 and ThrB30. The final synthetic yields were comparable to previous reports,18 demonstrating that the PEG-based connecting segments do not impair assembly of the linear precursor or its folding to native disulfide pairing. In Vitro Analysis of PEG Length. Each analog was tested for in vitro bioactivity in engineered human cells that overexpress the human insulin and IGF-1 receptors. The results are shown in Table 1 for binding and phosphorylation at the insulin receptor and phosphorylation at the IGF-1 receptor. The single chain insulin analogs possessing a full B-chain with PEG4, PEG8, and PEG16 linkers (peptides 3−5) all displayed low insulin activity (