Synthetic Advances in Insulin-like Peptides Enable ... - ACS Publications

Aug 3, 2017 - John Mayer,. † and Richard DiMarchi*,†,‡. †. Novo Nordisk Research Center Indianapolis, Indianapolis, Indiana 46241, United Stat...
6 downloads 10 Views 3MB Size
Article pubs.acs.org/accounts

Synthetic Advances in Insulin-like Peptides Enable Novel Bioactivity Published as part of the Accounts of Chemical Research special issue “Chemical Biology of Peptides”. Fa Liu,† Pengyun Li,† Vasily Gelfanov,† John Mayer,† and Richard DiMarchi*,†,‡ †

Novo Nordisk Research Center Indianapolis, Indianapolis, Indiana 46241, United States Department of Chemistry, Indiana University, Bloomington, Indiana 47405, United States



CONSPECTUS: Insulin is a miraculous hormone that has served a seminal role in the treatment of insulin-dependent diabetes for nearly a century. Insulin resides within in a superfamily of structurally related peptides that are distinguished by three invariant disulfide bonds that anchor the three-dimensional conformation of the hormone. The additional family members include the insulin-like growth factors (IGF) and the relaxin-related set of peptides that includes the so-called insulin-like peptides. Advances in peptide chemistry and rDNA-based synthesis have enabled the preparation of multiple insulin analogues. The translation of these methods from insulin to related peptides has presented unique challenges that pertain to differing biophysical properties and unique amino acid compositions. This Account presents a historical context for the advances in the chemical synthesis of insulin and the related peptides, with division into two general categories where disulfide bond formation is facilitated by native conformational folding or alternatively orthogonal chemical reactivity. The inherent differences in biophysical properties of insulin-like peptides, and in particular within synthetic intermediates, have constituted a central limitation to achieving high yield synthesis of properly folded peptides. Various synthetic approaches have been advanced in the past decade to successfully address this challenge. The use of chemical ligation and metastable amide bond surrogates are two of the more important synthetic advances in the preparation of high quality synthetic precursors to high potency peptides. The discovery and application of biomimetic connecting peptides simplifies proper disulfide formation and the subsequent traceless removal by chemical methods dramatically simplifies the total synthesis of virtually any two-chain insulin-like peptide. We report the application of these higher synthetic yield methodologies to the preparation of insulin-like peptides in support of exploratory in vivo studies requiring a large quantity of peptide. Tangentially, we demonstrate the use of these methods to study the relative importance of the IGF-1 connecting peptide to its biological activity. We report the translation of these finding in search of an insulin analog that might be comparably enhanced by a suitable connecting peptide for interaction with the insulin receptor, as occurs with IGF-1 and its receptor. The results identify a unique receptor site in the IGF1 receptor from which this enhancement derives. The selective substitution of this specific IGF-1 receptor sequence into the homologous site in the insulin receptor generated a chimeric receptor that was equally capable of signaling with insulin or IGF-1. This novel receptor proved to enhance the potency of lower affinity insulin ligands when they were supplemented with the IGF-1 connecting peptide that similarly enhanced IGF-1 activity at its receptor. The chimeric insulin receptor demonstrated no further enhancement of potency for native insulin when it was similarly prepared as a single-chain analogue with a native IGF-1 connecting peptide. These results suggest a more highly evolved insulin receptor structure where the requirement for an additional structural element to achieve high potency interaction as demonstrated for IGF-1 is no longer required.



therapy.2 Parallel advances in materials science, glucose sensors, and counter-regulatory hormones have accelerated the realization of closed-loop, bihormonal therapy with continuous glucose monitoring.3

INTRODUCTION

Insulin as a Therapeutic

The advent of rDNA biosynthesis provided a virtually unlimited supply of human insulin while the introduction of structural analogues enabled more precise prandial and basal control leading to improved disease outcomes and enhanced patient compliance. More recently, the emphasis has shifted to minimizing insulin dosage through combination therapy with GLP-1 based agents,1 a regimen well-suited for the treatment of T2D where obesity is a predominant factor. T1D insulin treatment is more fragile and complicates therapy to normalize glucose normalization, without increasing the risk of lifethreatening hypoglycemia. These needs have stimulated pursuit of polypharmacy for improved body weight management in T2D and the search for glucose-responsive insulins for T1D © 2017 American Chemical Society

Insulin as a Member of a Larger, Less Well-Characterized Superfamily

Insulin is the most prominent member of the insulin superfamily, which includes the insulin-like growth factors (IGF-1 and IGF-2) and seven members of the relaxin subfamily (relaxins 1, 2, and 3; insulin-like peptides INSL-3, 4, 5, and 6) (Figure 1). Members of the insulin superfamily share a unique, structural motif of six invariant cysteine residues that constitute Received: May 5, 2017 Published: August 3, 2017 1855

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 1. Human insulin family peptide sequences. Adapted with permission from ref 10. Copyright 2016 Wiley.

methods, and biomimetic folding promoted by single-chain intermediates.6−8,11−14 The second employs orthogonal cysteine protection to sequentially direct the formation of each individual disulfide.15−20

one intrachain and two interchain disulfide bonds. The insulinlike growth factors (IGF-1 and IGF-2) occur as single-chain linear peptides while the remaining members are processed to heterodimers in their mature forms. Analysis of genomic transcripts in 46 vertebrate and invertebrate species have identified the presence of insulin superfamily ligands, suggesting a broad distribution and essential roles throughout the animal kingdom.4 A number of peptides within the insulin superfamily are not readily accessible by biosynthetic and chemical approaches, limiting assessment of their biological function. Insulin-specific methodologies have been adapted to other members of this family with limited success. These challenges have stimulated creative efforts to achieve efficient synthesis of other insulin superfamily members. This Account highlights recent advances in this area, as well as structure− activity relationship in the previously unexplored connectingpeptide.

Chain Combination

The chief obstacle encountered in the initial chemical synthesis of insulin involved the assembly of the individual A- and Bchains which was first achieved by classical solution-phase peptide synthesis, with fragment condensation. The second obstacle pertained to proper connection of the three native disulfide bonds which was achieved through the elegant chain combination method.21 Du and colleagues later refined the conditions to a point where as much as 70% yield could be achieved.22 This led to the first reports in the total chemical synthesis of insulin by three independent teams.6−9 These reports represented a remarkable milestone in the history of peptide chemistry, but the overall yield was far less than 1%, largely a function of inefficient chain assembly. The application of solid-phase chemistry by Merrifield significantly improved the yield in A and B chain synthesis to 20−40%, and shortened the preparation time from months to days.11 Comparatively, the first synthesis of porcine relaxin was reported in 1981 and utilized the insulin chain combination method established nearly two decades earlier.23 Similar strategies were later employed in the preparation of human relaxin-124 and human relaxin-2.25 The low initial yields in relaxin synthesis were attributed to the poor solubility of its Bchain which was purportedly enhanced through addition of organic solvents, or denaturing agents when added to the folding buffer.26 Separately, B-chain truncation (1−25), its



CHEMICAL SYNTHESIS OF INSULIN-FAMILY PEPTIDES Initial attempts at a total chemical synthesis of insulin started shortly after the landmark sequence elucidation by Sanger and colleagues in 19555 with the first successes reported in mid 1960s.6−8 Structure−activity relationship studies of insulin required substantial refinement of these methods, as well as additional independent means to prepare specific analogues.9,10 The reported syntheses can be categorized according to the specific method used in the formation of the disulfide bonds. The first category relies upon native conformational forces, and it includes the initial unassisted random two-chain combination 1856

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 2. Biomimetic synthesis of insulin superfamily peptides by oxime ligation and traceless “C-peptide” chemical excision.37,38

single-chain form to native two-chain, relaxin-2 was achieved by sequential treatment with Asp-N and trypsin proteases.31 Similar biosynthetic methods were used to prepare INSL-3,32 INSL-5,33 relaxin-334 and INSL-6,35 although each with a slightly different mini “C-peptide” in the single-chain precursor form. The Kent group utilized a mini-PEG tether as a surrogate for a connecting peptide, and employed an oxime-based ligation to connect the A-chain N-terminus to the B-chain C-terminus.12 The linker was enzymatically excised after disulfide formation to afford insulin lispro at 15% combined yield, over these last two steps. The same group published a second route utilizing ester bond formation between the side chains of GluA4 and ThrB30 as the chemical tether. This SCI precursor had been prepared in 70% overall yield by native chemical ligation (NCL) of a PheB1-ValB18 peptide thioester and a CysB19AsnA21 peptide, with the latter bearing the ester bond linkage.36 Our group37,38 recently established a procedure which uses oxime-ligation to join purified A and B-chains, and an ester bond to link the ThrB30 carboxyl and a B-chain aminooxy-handle. A dipeptide was inserted between GlyA1 and an aminooxy-handle to the N-terminus of the A-chain. Following oxidative folding in aqueous alkaline buffer, the dipeptide was rearranged through diketopiperazine (DKP) cleavage within 2 h at ∼60 °C in a neutral pH buffer. The ester bond was subsequently saponified at 4 °C and pH 11. Excision of the A-B chain linker through two chemical steps liberated active two-chain insulin, and the linker as a synthetic byproduct. Both steps proceeded cleanly and efficiently in a combined yield of 45% (Figure 2). We discovered that in analogous fashion to the DesDi precursor the folding of this linear SCI tolerated destabilizing mutations, as exemplified by successful folding of the reportedly resistant ValA16 analogue.39 The

extension (1−33), and chemical oxidation of MetB25 to the sulfoxide form Met(O) were reported to improve solubility and subsequently total synthetic yields.24,25,27 The B-chaintruncated or extended relaxin-2 analogues demonstrated similar biological activity as the native hormone, while the Met(O) analogue could be readily reduced to native sequence by ammonium iodide in TFA.25 Chain Combination through Intramolecular Folding

The variable yields realized through chain combination9 prompted the exploration of single-chain, biomimetic routes as inspired by the efficient folding characteristics of native proinsulin.28 The earliest biomimetic precursors included miniproinsulin29 and porcine insulin precursor, PIP.30 The former represents a 50-residue desB30 single-chain intermediate linking LysB29 directly to GlyA1 through a single amide bond, while the latter employs a non-native Ala-Ala-Lys linker. Both precursors fold in high yield and the resulting single-chain insulin (SCI) intermediates undergo efficient proteolytic conversion to fully active two-chain insulin.29 Both of these SCI analogues have been utilized in rDNA-biosynthetic commercial insulin manufacture. Recently, Tofteng and colleagues disclosed an SCI-precursor linked by a Gly-Glu-Glu-Glu-Lys “C-peptide” surrogate sequence13 which produced an overall synthetic yield of 12%. Efforts in our laboratory focused on a synthetic 49-residue SCI precursor termed “DesDi” derived by the deletion of the B28 and B30 residues, with direct amide bond linkage of GlyA1 and LysB29 folded with high efficiency and tolerated conformationdestabilizing mutations.14 The preparation of other insulin superfamily peptides using an SCI intermediate has been reported. The 102-residue C-peptide of relaxin-2 was replaced with a shorter, corresponding sequence from IGF-1 enabling efficient folding and disulfide-formation. The conversion of the 1857

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 3. Directed-disulfide bond formation-based synthetic routes to relaxin-2, bombyxin, and human insulin.16,42,43

chemical reactivity of orthogonally protected cysteine pairs, a strategy that is generally compatible with any sequence mutation. Consequently, this approach represents a more reliable synthetic method in the preparation of novel, nonnative analogues. The first reported synthesis of insulin by this approach was the classic 1974 solution phase synthesis reported by the Sieber group in which the disulfides were formed regiospecifically by using iodine oxidation, or sulfenyl thiocarbonate activation.15,41 The first synthesis of relaxin-2 by a directed disulfide bond formation route was reported by Büllesbach and Schwabe42 using Fmoc/tBu-chemistry for the A-chain and a Boc/Bzl protocol for the B-chain. The intra A-chain disulfide bond was formed via air oxidation. A 4-MeBzl group was used to protect CysA24, and it was removed with HF-treatment prior to (nitropyridinyl)thiol-directed thiolysis with CysB23. Iodine oxidative treatment was applied to establish the last disulfide bond from Cys(Acm) protected residues. A final alkaline treatment and reduction by ammonium iodide/TFA deprotected the Trp(CHO) and Met(O) residues to yield native relaxin-2 (Figure 3).

newly developed method was successfully applied to the synthesis of relaxin-2 and INSL-5.38 Unlike the conserved GlyA1 residue of insulin, both relaxin-2 and INSL-5 possess a pyroglutamic acid at the A1 position. Glutamine was used at A1 for attachment of the aminooxy-handle with anticipation of its conversion to pyroglutamate at a subsequent point in the synthesis. The oxime-mediated ligation, and the subsequent folding proceeded with equal efficiency as experienced in the analogous synthesis of insulin. The pyroglutamate at A1 formed concomitantly with DKP cyclization and linker detachment. Overall, the concerted DKP cyclization and saponification provided the native hormones at 10−20% yield, as calculated from purified A or B-chains. Directed Disulfide Bond Formation by Orthogonal Cysteine Protection

The biomimetic strategy enabled by conformational factors can often provide less than satisfactory yields, particularly in cases where a specific mutation disrupts the native folding of the single-chain precursor,40 a limitation that effectively restricts the scope of structure function studies.9 Alternatively directed disulfide bond formation can be achieved unambiguously by 1858

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 4. Insulin synthesis employing structural modifications to enhance aqueous solubility. (A) Isoelectric point adjustment of A-chain,17 (B) isoacyl bond surrogate in A-chain and B-chain,20 and (C) isoacyl bond surrogate in A- and B-chains as route to synthesis of a four disulfide bonded insulin.50

The synthesis of bombyxin, a peptide closely associated with the insulin superfamily, was concurrently reported by the Suzuki group. Each of the individual bombyxin chains were assembled by Fmoc/tBu- solid-phase chemistry using orthogonal Cys protection (Acm, Trt, tBu), which facilitated threestep disulfide bond formation in an analogous fashion to relaxin-2.43 The key difference distinguishing the Suzuki and Büllesbach routes was the use of Cys(tBu) in the former, and Cys(MeBzl) in the latter. The conversion of Cys(tBu) to Cys(SPy) was achieved in anhydrous TFA, in the presence of dithiodipyridine and TFMSA (Figure 3). This strategy was readily adopted and utilized in a number of subsequent relaxin syntheses.44,45 The reported chemical syntheses of the insulinlike peptides have in general also followed Suzuki’s directed disulfide bond formation strategy, including INSL-3,46 INSL4,47 and INSL-5.48 Shortly after publication of the bombyxin and relaxin-2 syntheses the Kiso group reported a solid-phase synthesis of human insulin using directed disulfide bond formation16 (Figure 3). The first disulfide bond was established at CysA20 and CysB19 by thiopyridyl-directed thiolysis. The resulting heterodimer was treated with iodine to oxidatively convert Cys(Acm)A7 and Cys(Acm)B7 to the second disulfide. The A-intrachain disulfide bond was constructed using Kiso’s silyl chloride-sulfoxide method. Despite directed heterodimer formation the total yield was no more than 1% when calculated from starting resin. Several solutions have been advanced to address the poor recovery of insulin A-chain and the A−B heterodimers during intermediate preparative chromatography steps. These include one-pot routes that avoid the purification step, as well as

methods that enhance A-chain solubility. The latter approach includes the use of temporary solubilizing tags, and the isoacylpeptide strategy. The Wade group proposed a pentalysine tag as a C-terminal A-chain extension conjugated via an ester bond,17 which significantly improved the solubility and chromatographic handling of the synthetic intermediates (Figure 4). The tag was released in the final step by saponification to yield insulin glargine. More recently, Liu et al. utilized an isoacyl approach to insulin synthesis. The superior biophysical properties of the resulting isoacyl A and Bchains dramatically improved the yield in all steps, including individual chain assembly, purification and subsequent disulfide bond formation. The O-to-N acyl shift occurs in the last synthetic step at neutral pH and human insulin is obtained in the highest synthetic yield reported to date, 24% based on Achain resin substitution (Figure 4).20 Iodine-Free Synthesis

Iodine oxidation has been routinely used in the synthesis of insulin and superfamily peptides to convert Acm-protected cysteines to one of the three disulfide bonds.9,49 However, this strong oxidative reagent is inherently limiting due its potential to modify methionine and tryptophan. Unlike insulin this offtarget reactivity is a limitation in the synthesis of insulin-like peptides where these oxidation-sensitive amino acids are common, and more generally it prohibits the inclusion of any similarly sensitive amino acid mimetics. Directed synthesis that is independent of iodine would expand the range of multiplecysteine containing peptides that could be prepared in high yield to support in vivo studies. 1859

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 5. Insulin superfamily peptide synthesis employing disulfide formation by iodine-free routes18,19,51,52

Figure 6. Insulin analogue 010 in vitro activity as measured in HEK293 cells that overexpress the human insulin A-isoform receptor (A), human insulin-B isoform receptor (B), or the human IGF-1 receptor (C), with the amino acid sequence of 010 shown at the top of the figure (D).

plished by the formation of the first two disulfide bonds via (nitro-pyridinyl)thiol-directed thiolysis, while the last disulfide was formed by treatment of a Cys(Mob)/Cys(Mob) sulfoxide pair in TFA and TFMSA, in the presence of diphenylsulfide as a soft base, which proved critical to achieving a high yield. Recently, Liu et al. reported the use of penicillin G acylase (PGA)-labile phenylacetamidomethyl (Phacm) group to

In a recent synthesis of INSL-6, we reported that the addition of tryptophan to the iodine-oxidation reaction minimized oxidation of this residue, without impairing the conversion of the disulfide from Cys(Acm)50 (Figure 5). Separately, we explored the utility of Yajima’s sulfoxide-directed disulfide bond formation51 in the synthesis of insulin superfamily peptides. The synthesis of INSL-6 was accom1860

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 7. Relative bioactivity of insulin analogue 010 with Ala-substitution in the C-peptide at the human insulin A-isoform receptor and human IGF-1 receptor (A), and relative binding affinity at the two insulin receptor isoforms (B). The amino acid sequence of 010 is shown at the top of the figure.



protect CysA7 and B719 in the successful preparation of a porcine insulin analogue. In the presence of Ellman’s reagent, the first liberated thiol was trapped as an activated intermediate to facilitate subsequent thiolysis-directed disulfide bond formation as the second Phacm was removed by PGA, to furnish the last disulfide in 21% yield. This method was subsequently applied to the synthesis of relaxin-2. In the initial synthesis, the use of Ellman’s reagent produced a significant amount of free A- and B-chains as unwanted side-products, but its replacement with bis(5-(2-methoxyethoxy)-2-pyrimidinyl disulfide (BMPD) afforded relaxin-2 in 5.8% yield, as calculated from B-chain resin substitution (Figure 5).52 An alternative method has been reported by the Wade group, where the photolabile 2-nitroveratryl group (oNv) was employed for side-chain protection of CysA7. The photolytic unmasking of the Cys7 thiol was immediately followed with disulfide bond formation as directed by a thiopyridyl activated cysteine, affording human insulin in high yield (Figure 5).18

APPLICATION OF SYNTHETIC CHEMISTRY TO STUDY THE COMPARATIVE RELATIONSHIP OF INSULIN AND IGF-1 Insulin and IGF-1 share approximately 50% homology in their respective A- and B-chains (Figure 1). The comparative analysis of these two hormones inspired the design of lispro, insulin.57 The most striking difference among the two biologically active peptides is the fact that insulin functions as a two-chain peptide where IGF-1 maintains its linear structure with a 12 residue connecting peptide. Use of the IGF-1 derived connecting peptide to similarly link the insulin A- and B-chains is illustrated in Figure 6. We refer to it as 010 where the 0 is reflective of the insulin A- and B-chain sequences and the 1 being IGF-1 derived. The peptide is oxidatively folded to the correct disulfides using conditions previously reported for in the biosynthesis of IGFs, and the yield for all insulin analogues studied was in excess of 70%.58 When studied for its ability to biochemically signal in cells engineered to overexpress the IGF1 receptor, or one of the insulin receptor isoforms it is clear that 010 possesses hybrid properties that reflect the higher potency that is inherent to each of the native hormones. It is a fully balanced agonist at each of the three receptors and equally potent as the native hormone. To investigate the basis of this enhanced activity, we completed an alanine scan within the Cpeptide and studied the ability of each of these peptides to function at one of the three receptors. The comparative performance in signaling at an insulin receptor relative to the related IGF-1 receptor is shown in Figure 7A. The native hormones demonstrate subnanomolar potency and selectivity that is approximately 100-fold enhanced with respect to their cognate receptors. The chimeric peptide 010 in these determinations is nearly balanced for the two receptors and slightly less potent than the native hormones at their individual receptors. The activity within the set of alanine mutants at the insulin receptor is relatively constant as might be expected given that there is no connecting peptide in native insulin. Nonetheless, there are subtle differences, most notably enhanced potency at the fourth position and lesser potency at positon seven. The IGF-1 receptor interaction within the same set of peptides demonstrates reduced potency. Most of the analogues exhibit a subtle reduction in potency, but positons two, three, and seven are notably lowered in IGF-1

A Four Disulfide Bond Insulin

Recently reported analogues with four-disulfide bonds are a manifestation of the progress made in the synthesis of insulin and related peptides. These analogues, which were prepared by biosynthetic methods, exhibit increased physical stability and protease resistance.53 While yeast-based expression and folding worked well in the case of certain analogues, in other instances these methods failed to produce other analogues in sufficient quantity to support biological study. In contrast, a newly established four-step, sequentially directed chemical route from our group appears more broadly applicable and capable in generating peptides that could not be accessed biosynthetically.54 The synthetic scheme builds upon the Liu method20 of using the isoacyl dipeptide units20 and the DMSO/TFA oxidation of two Cys(tBu) residues51 to furnish the fourth disulfide bond. The method was used to achieve the first synthesis of the recently reported snail insulin G2,55 a fourdisulfide insulin-like peptide with little sequence homology to human insulin (Figure 4). It is worth nothing that a distantly related two-chain four-disulfide-bond insulin-like peptide, the crustacean androgenic gland hormone, has been prepared using three different orthogonal reactions to produce a two-chain hormone with four disulfide bonds.56 1861

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

Figure 8. Chimeric insulin, IGF-1 receptor (A), where the specific insulin receptor residues 259−284 were replaced with IGF-1 residues 252−273 (B), the three-dimensional location of this specific peptide within the cysteine-region of the receptor (C), and the relative biochemical signaling (calculated EC50) for full agonism of three single chain insulin analogues, at the insulin receptor B isoform relative to the chimeric insulin receptor with IGF-1 residues 253−274.

the potency of 010. This suggests that native insulin has been sufficiently optimized for recognition at the insulin receptor such that further enhancement is unnecessary and serves little useful purpose. To further interrogate this observation, we prepared two 010 analogues that possess modifications that serve to reduce potency when introduced in native insulin (without the connecting peptide of 010). The substitution of the native B24Phe with Ser or B16,17 Tyr-Leu with Gln-Phe reduced the potency at the insulin receptor by more than 30fold. These same analogues proved to be of comparable potency to native insulin and 010 at the chimeric insulin receptor, demonstrating that this additional binding site can serve to sizably potentiate the potency of weaker insulin ligands. This provides a plausible explanation for IGF-1 maintaining a connecting peptide to potentiate its interaction with its receptor something that is unnecessary for insulin, and thus a basis for a two-chain ligand.

activity and serve to distort the balanced potency observed in 010 to a more selective insulin bias, but not to the extent of native insulin. A comparable analysis of relative binding at the two insulin receptor isoforms is informative with all of the alanine analogues of 010 behaving in similar, but distinct fashion (Figure 7B). Most notably alanine substitution at positions one and eight generated 010 analogues that were biased toward isoform A, while the opposite was observed for substitution in the last three residues. Whether these modest changes can be amplified in substituting something other than alanine, or through simultaneous changes at these same sites is a subject of continued investigation. Having determined that the IGF-1 receptor is more sensitive to changes in the connecting peptide sequence we hypothesized that there must be a site in the IGF-1 receptor that is different from that in the analogous insulin receptor. Given the relative cationic nature of the IGF-1 connecting peptide, we searched the alignment of the IGF-1 receptor for a putative binding site populated with anionic amino acids that did not exist in the insulin receptor. This analysis identified a twentytwo residue peptide flanked by cysteine residues in the cysteinerich region of the IGF-1 receptor.59 The homologous region in insulin was slightly longer and strikingly of inverse ionic character, with multiple lysine, arginine and histidine residues (Figure 8A,B). A chimeric insulin receptor was assembled that substituted the putative binding region for the IGF-1 connecting peptide (Figure 8C). Its biochemical character was studied and where the native insulin receptor demonstrated a nearly 100x preference for insulin relative to IGF-1 the chimeric receptor displayed no change in the ability to recognize insulin, but a more than 10x increase for IGF-1. While this receptor site clearly enhanced the recognition for the connecting peptide rendering IGF-1 a superior ligand relative to the native insulin receptor, it did not appreciably enhance



CONCLUSION The advances in synthetic chemistry that we detail in this Account provide a foundation for the chemical preparation within this family of peptides in sufficient quantity to support thorough biological study. Through recruitment and refinement of an IGF-1 based C-peptide, we report single chain insulin analogues that demonstrate balanced and full potency at the IGF1 and insulin receptor isoforms. We also show how sitespecific substitution with alanine can be used to selectively embellish receptor-based, intracellular signaling through these insulin-related receptors. In particular, the identification of peptides that exhibit enhanced activity at one or the other insulin receptor isoform provides a basis to pharmacologically interrogate the relative importance of these two receptors. Finally, these fully potent and balanced insulin-IGF1 coagonists 1862

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research have led to identification of a specific site within these related receptors that serves to enhance the activity of single-chain insulin-based agonists. We close this report with great optimism for the future and anticipation of what the second century in insulin chemistry and biology will bring. For example, we look to answers for why nature possesses two insulin receptor isoforms. How do two receptors as similar as IGF1 and insulin lead to such dramatically different physiological consequences when there is such congruence in their intracellular signaling? Finally, can we successfully harness our academic discoveries to provide more efficacious and safer insulin-based therapies?



Chemistry. He is also a 2014 inductee to the National Inventors Hall of Fame.



REFERENCES

(1) Heppner, K. M.; Perez-Tilve, D. GLP-1 based therapeutics: simultaneously combating T2DM and obesity. Front. Neurosci. 2015, 9, 92. (2) Zaykov, A. N.; Mayer, J. P.; DiMarchi, R. D. Pursuit of a perfect insulin. Nat. Rev. Drug Discovery 2016, 15, 425−439. (3) Yang, J.; Cao, Z. Glucose-responsive insulin release: Analysis of mechanisms, formulations, and evaluation criteria. J. Controlled Release 2017, DOI: 10.1016/j.jconrel.2017.01.043. (4) Shabanpoor, F.; Separovic, F.; Wade, J. D. The human insulin superfamily of polypeptide hormones. Vitam. Horm. 2009, 80, 1−31. (5) Ryle, A. P.; Sanger, F.; Smith, L. F.; Kitai, R. Disulphide bonds of insulin. Biochem. J. 1955, 60, 541−556. (6) Meienhofer, J.; Zahn, H.; Schnabel, E.; Bremer, H.; Zabel, R.; Sroka, W.; Okuda, T.; Brinkhoff, O.; Brandenburg, D.; Klostermeyer, H. Synthese der insulinketten und ihre kombination zu insulinaktiven praparaten. Z. Naturforsch., B: J. Chem. Sci. 1963, 18, 1120−1121. (7) Kung, Y. T.; Du, Y. C.; Huang, W. T.; Chen, C. C.; Ke, L. T.; Hu, S. C.; Jiang, R. Q.; Chu, S. Q.; Niu, C. I.; Hsu, J. Z.; Chang, W. C.; Cheng, L. L.; Li, H. S.; Wang, Y.; Loh, T. P.; Chi, A. H.; Li, C. H.; Shi, P. T.; Yieh, Y. H.; Tang, K. L.; Hsing, C. Y. Total synthesis of crystalline bovine insulin. Sci. Sin. 1965, 14, 1710−1716. (8) Katsoyannis, P. G.; Tometsko, A.; Zalut, C. Insulin peptides 0.12: human insulin generation by combination of synthetic A and B chains. J. Am. Chem. Soc. 1966, 88, 166−167. (9) Mayer, J. P.; Zhang, F.; DiMarchi, R. D. Insulin structure and function. Biopolymers 2007, 88, 687−713. (10) Liu, F.; Zaykov, A. N.; Levy, J. J.; Di Marchi, R. D.; Mayer, J. P. Chemical synthesis of peptides within the insulin superfamily. J. Pept. Sci. 2016, 22, 260−270. (11) Marglin, A.; Merrifield, R. B. Synthesis of bovine insulin by solid phase method. J. Am. Chem. Soc. 1966, 88, 5051−5052. (12) Sohma, Y.; Kent, S. B. H. Biomimetic synthesis of lispro insulin via a chemically synthesized ″mini-proinsulin″ prepared by oximeforming ligation. J. Am. Chem. Soc. 2009, 131, 16313−16318. (13) Tofteng, A. P.; Jensen, K. J.; Schaeffer, L.; Hoeg-Jensen, T. Total synthesis of desB30 insulin analogues by biomimetic folding of singlechain precursors. ChemBioChem 2008, 9, 2989−2996. (14) Zaykov, A. N.; Mayer, J. P.; Gelfanov, V. M.; DiMarchi, R. D. Chemical synthesis of insulin analogs through a novel precursor. ACS Chem. Biol. 2014, 9, 683−691. (15) Sieber, P.; Kamber, B.; Hartmann, A.; Johl, A.; Riniker, B.; Rittel, W. Total synthesis of human insulin by direct formation of disulfide bonds-preliminary report. Helv. Chim. Acta 1974, 57, 2617− 2621. (16) Akaji, K.; Fujino, K.; Tatsumi, T.; Kiso, Y. total synthesis of human insulin by regioselective disulfide formation using the silyl chloride sulfoxide method. J. Am. Chem. Soc. 1993, 115, 11384−11392. (17) Hossain, M. A.; Belgi, A.; Lin, F.; Zhang, S.; Shabanpoor, F.; Chan, L.; Belyea, C.; Truong, H.-T.; Blair, A. R.; Andrikopoulos, S.; Tregear, G. W.; Wade, J. D. Use of a temporary ″solubilizing″ peptide tag for the Fmoc solid-phase synthesis of human insulin glargine via use of regioselective disulfide bond formation. Bioconjugate Chem. 2009, 20, 1390−1396. (18) Karas, J. A.; Scanlon, D. B.; Forbes, B. E.; Vetter, I.; Lewis, R. J.; Gardiner, J.; Separovic, F.; Wade, J. D.; Hossain, M. A. 2-Nitroveratryl as a photocleavable thiol-protecting group for directed disulfide bond formation in the chemical synthesis of insulin. Chem. - Eur. J. 2014, 20, 9549−9552. (19) Liu, F.; Liu, Q.; Mezo, A. R. An iodine-free and directeddisulfide-bond-forming route to insulin analogues. Org. Lett. 2014, 16, 3126−3129. (20) Liu, F.; Luo, E. Y.; Flora, D. B.; Mezo, A. R. A synthetic route to human insulin using isoacyl peptides. Angew. Chem., Int. Ed. 2014, 53, 3983−3987.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Fa Liu: 0000-0002-2909-9185 Richard DiMarchi: 0000-0003-0220-4085 Notes

The authors declare the following competing financial interest(s): Employment at Novo Nordisk and intellectual property at Indiana University. Biographies Fa Liu is currently the Director of Chemistry at Novo-Nordisk, Seattle Research Center. Dr. Liu received his Ph.D. from the Shanghai Institute of Organic Chemistry in 2004 followed by studies at the National Cancer Institute (2004−2009). He started his pharmaceutical career at Lilly Research Laboratories where after 5 years he joined Calibrium, and subsequently Novo-Nordisk in 2016. Pengyun Li is currently research scientist at Novo-Nordisk Company. He received his Ph.D. from Tsinghua University in 2003 and studied as a postdoctoral fellow at the University of Chicago (2004−2005). He subsequently worked as a research scientist at Indiana University (2005−2016) before joining Novo-Nordisk. Vasily M. Gelfanov is a Principal Research Scientist at Novo-Nordisk. He received his Ph.D. from the Shemyakin Institute of Bioorganic Chemistry, Russian Academy of Sciences, Moscow, Russian Federation (1992). He completed postdoctoral studies at the Institute of Molecular Biology Academia Sinica, Taipei, Taiwan (1993−1996). He has more than 20 years of research experience in biotechnology with a particular emphasis on cellular biochemistry. Prior to joining Novo-Nordisk, he held a faculty position at Indiana University (2004− 2016). John P. Mayer is a Principal Research Scientist at Novo-Nordisk Research Laboratories in Indianapolis. He received his Ph.D. from Purdue University in 1987 followed by postdoctoral work at Eli Lilly Research Laboratories. He was employed at Amgen (1992−1998) and Lilly (1998−2012) and held a faculty position at Indiana University (2012−2016). He served as member of the American Peptide Society governing board of directors. Richard DiMarchi is a Distinguished Professor of Biochemistry and Gill Chair in Biomolecular Sciences at Indiana University. He is also Site Director of the Novo-Nordisk Indianapolis Research Center in Indianapolis. He is Chairman of the Peptide Therapeutics Foundation and a member of the National Academy of Medicine. Dr. DiMarchi is the recipient of numerous awards including the 2011 Merrifield Award for Career Contributions in Peptide Sciences, and the 2016 ACS Alfred Burger Award for Career Achievements in Medicinal 1863

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research

″nonfoldable″ insulin imparied folding efficiency despite native activity. J. Biol. Chem. 2009, 284, 35259−35272. (40) Guo, Z. Y.; Wang, S.; Tang, Y. H.; Feng, Y. M. Mutagenesis of the three conserved valine residues: consequence on the foldability of insulin. Biochim. Biophys. Acta, Proteins Proteomics 2004, 1699, 103− 109. (41) Sieber, P.; Kamber, B.; Hartmann, A.; Joehl, A.; Riniker, B.; Rittel, W. Total synthesis of human insulin. IV. Description of the final steps. Helv. Chim. Acta 1977, 60, 27−37. (42) Büllesbach, E. E.; Schwabe, C. Total synthesis of human relaxin and human relaxin derivatives by solid-phase peptide-synthesis and site-directed chain combination. J. Biol. Chem. 1991, 266, 10754− 10761. (43) Maruyama, K.; Nagasawa, H.; Isogai, A.; Tamura, S.; Ishizaki, H.; Suzuki, A. Structure and synthesis of bombyxin-IV, a brain peptide of the silkworm Bombyx mori. Pept. Chem. 1990, 27, 251−256. (44) Bathgate, R. A. D.; Lin, F.; Hanson, N. F.; Otvos, L.; Guidolin, A.; Giannakis, C.; Bastiras, S.; Layfield, S. L.; Ferraro, T.; Ma, S.; Zhao, C. X.; Gundlach, A. L.; Samuel, C. S.; Tregear, G. W.; Wade, J. D. Relaxin-3: Improved synthesis strategy and demonstration of its highaffinity interaction with the relaxin receptor LGR7 both in vitro and in vivo. Biochemistry 2006, 45, 1043−1053. (45) Hossain, M. A.; Lin, F.; Zhang, S.; Ferraro, T.; Bathgate, R. A.; Tregear, G. W.; Wade, J. D. Regioselective disulfide solid phase synthesis, chemical characterization and in vitro receptor binding activity of equine relaxin. Int. J. Pept. Res. Ther. 2006, 12, 211−215. (46) Del Borgo, M. P.; Hughes, R. A.; Bathgate, R. A. D.; Lin, F.; Kawamura, K.; Wade, J. D. Analogs of insulin-like peptide 3 (INSL3) B-chain are LGR8 antagonists in vitro and in vivo. J. Biol. Chem. 2006, 281, 13068−13074. (47) Büllesbach, E. E.; Schwabe, C. Synthesis and conformational analysis of the insulin-like 4 gene product. J. Pept. Res. 2001, 57, 77− 83. (48) Akhter Hossain, M.; Bathgate, R. A. D.; Kong, C. K.; Shabanpoor, F.; Zhang, S.; Haugaard-Jonsson, L. M.; Rosengren, K. J.; Tregear, G. W.; Wade, J. D. Synthesis, conformation, and activity of human insulin-like peptide 5 (INSL5). ChemBioChem 2008, 9, 1816− 1822. (49) Belgi, A.; Akhter Hossain, M.; Tregear, G. W.; Wade, J. D. The chemical synthesis of insulin: from the past to the present. Immunol., Endocr. Metab. Agents Med. Chem. 2011, 11, 40−47. (50) Wu, F.; Mayer, J. P.; Zaykov, A. N.; Zhang, F.; Liu, F.; DiMarchi, R. D. Chemical synthesis of human insulin-like peptide-6. Chem. - Eur. J. 2016, 22, 9777−9783. (51) Fujii, N.; Otaka, A.; Watanabe, T.; Arai, H.; Funakoshi, S.; Bessho, K.; Yajima, H. Sulfoxide-directed disulfide bond-forming reaction for the synthesis of cystine peptides. J. Chem. Soc., Chem. Commun. 1987, 1676−1678. (52) Yang, X.; Gelfanov, V. M.; Liu, F.; DiMarchi, R. D. Synthetic route to human relaxin-2 via iodine-free sequential disulfide bond formation. Org. Lett. 2016, 18, 5516−5519. (53) Vinther, T. N.; Norrman, M.; Ribel, U.; Huus, K.; Schlein, M.; Steensgaard, D. B.; Pedersen, T. A.; Pettersson, I.; Ludvigsen, S.; Kjeldsen, T.; Jensen, K. J.; Hubalek, F. Insulin analog with additional disulfide bond has increased stability and preserved activity. Protein Sci. 2013, 22, 296−305. (54) Wu, F.; Mayer, J. P.; Gelfanov, V. M.; Liu, F.; DiMarchi, R. D. Synthesis of four-disulfide insulin analogs via sequential disulfide bond formation. J. Org. Chem. 2017, 82, 3506−3512. (55) Safavi-Hemami, H.; Gajewiak, J.; Karanth, S.; Robinson, S. D.; Ueberheide, B.; Douglass, A. D.; Schlegel, A.; Imperial, J. S.; Watkins, M.; Bandyopadhyay, P. K.; Yandell, M.; Li, Q.; Purcell, A. W.; Norton, R. S.; Ellgaard, L.; Olivera, B. M. Specialized insulin is used for chemical warfare by fish-hunting cone snails. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 1743−1748. (56) Katayama, H.; Hojo, H.; Ohira, T.; Ishii, A.; Nozaki, T.; Goto, K.; Nakahara, Y.; Takahashi, T.; Hasegawa, Y.; Nagasawa, H.; Nakahara, Y. Biochemistry 2010, 49, 1798−1807.

(21) Dixon, G. H.; Wardlaw, A. C. Regeneration of insulin activity from the separated and inactive A and B chains. Nature (London, U. K.) 1960, 188, 721. (22) Du, Y.-T.; Jiang, R.-Q.; Tsou, C.-L. Conditions for successful resynthesis of insulin from its glycyl and phenylalanyl chains. Sci. Sin. (Engl. Ed.) 1965, 14, 229−236. (23) Tregear, G. W.; Fagan, C.; Reynolds, H.; Scanlon, D.; Jones, P.; Kemp, B.; Niall, H. D.; Du, Y. C. Porcine relaxin: synthesis and structure activity relationships. Pept. Synth., Struct., Funct., Proc. Am. Pept. Symp. 1981, 7th, 249−252. (24) Hudson, P.; Haley, J.; John, M.; Cronk, M.; Crawford, R.; Haralambidis, J.; Tregear, G.; Shine, J.; Niall, H. Structure of a genomic clone encoding biologically active human relaxin. Nature 1983, 301, 628−631. (25) Barlos, K. K.; Gatos, D.; Vasileiou, Z.; Barlos, K. An optimized chemical synthesis of human relaxin-2. J. Pept. Sci. 2010, 16, 200−211. (26) Hossain, M. A.; Wade, J. D. Synthetic relaxins. Curr. Opin. Chem. Biol. 2014, 22, 47−55. (27) Hudson, P.; John, M.; Crawford, R.; Haralambidis, J.; Scanlon, D.; Gorman, J.; Tregear, G.; Shine, J.; Niall, H. Relaxin gene expression in human ovaries and the predicted structure of a human preprorelaxin by analysis of cDNA clones. EMBO J. 1984, 3, 2333− 2339. (28) Steiner, D. F. Proinsulin and the biosynthesis of insulin. N. Engl. J. Med. 1969, 280, 1106−1113. (29) Markussen, J.; Jorgensen, K. H.; Sorensen, A. R.; Thim, L. Single chain des-(B30) insulin. Intramolecular crosslinking of insulin by trypsin catalyzed transpeptidation. Int. J. Pept. Protein Res. 1985, 26, 70−77. (30) Chang, S. G.; Kim, D. Y.; Choi, K. D.; Shin, J. M.; Shin, H. C. Human insulin production from a novel mini-proinsulin which has high receptor-binding activity. Biochem. J. 1998, 329, 631−635. (31) Vandlen, R.; Winslow, J.; Moffat, B.; Rinderknecht, E. Human relaxin: Purification, characterization and production of recombinant relaxins for structure function studies. Progress in Relaxin Research, International Congress on the Hormone Relaxin, 2nd, Adelaide, Mar. 6−12, 1995; Vol. 1994, pp 59−72. (32) Luo, X.; Bathgate, R. A. D.; Liu, Y.-L.; Shao, X.-X.; Wade, J. D.; Guo, Z.-Y. Recombinant expression of an insulin-like peptide 3 (INSL3) precursor and its enzymatic conversion to mature human INSL3. FEBS J. 2009, 276, 5203−5211. (33) Luo, X.; Bathgate, R. A. D.; Zhang, W.-J.; Liu, Y.-L.; Shao, X.-X.; Wade, J. D.; Guo, Z.-Y. Design and recombinant expression of insulinlike peptide 5 precursors and the preparation of mature human INSL5. Amino Acids 2010, 39, 1343−1352. (34) Luo, X.; Liu, Y.-L.; Layfield, S.; Shao, X.-X.; Bathgate, R. A. D.; Wade, J. D.; Guo, Z.-Y. A simple approach for the preparation of mature human relaxin-3. Peptides 2010, 31, 2083−2088. (35) Luo, X.; Huang, C.-H.; Shao, X.-X.; Guo, Z.-Y. Design, recombinant expression and in vitro maturation of human insulin-like peptide 6 and a biotin-labeled analogue. Process Biochem. (Oxford, U. K.) 2011, 46, 1243−1247. (36) Sohma, Y.; Hua, Q.-X.; Whittaker, J.; Weiss, M. A.; Kent, S. B. H. Design and folding of Glu(A4)(O(beta)Thr(B30)) insulin (″ester Insulin″): a minimal proinsulin surrogate that can be chemically converted into human insulin. Angew. Chem., Int. Ed. 2010, 49, 5489− 5493. (37) Thalluri, K.; Kou, B.; Gelfanov, V.; Mayer, J. P.; Liu, F.; DiMarchi, R. D. Biomimetic synthesis of insulin enabled by oxime ligation and traceless ″C-Peptide″ chemical excision. Org. Lett. 2017, 19, 706−709. (38) Thalluri, K.; Kou, B.; Yang, X.; Zaykov, A. N.; Mayer, J. P.; Gelfanov, V. M.; Liu, F.; DiMarchi, R. D. Synthesis of relaxin-2 and insulin-like peptide 5 enabled by novel tethering and traceless chemical excision. J. Pept. Sci. 2017, 23, 455−465. (39) Liu, M.; Wan, Z.-l.; Chu, Y.-C.; Aladdin, H.; Klaproth, B.; Choquette, M.; Hua, Q.-X.; Mackin, R. B.; Rao, J. S.; De Meyts, P.; Katsoyannis, P. G.; Arvan, P.; Weiss, M. A. Crystal structure of a 1864

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865

Article

Accounts of Chemical Research (57) Di Marchi, R. D.; Mayer, J. P.; Fan, L.; Brems, D. N.; Frank, B. H.; Green, L. K.; Hoffmann, J. A.; Howey, D. C.; Long, H. B.; et al. Synthesis of a fast-acting insulin based on structural homology with insulin-like growth factor I. Pept.: Chem. Biol. Proc. Am. Pept. Defense Technical Information Center: Fort Belvoir 1992, 12th, 26−28. (58) Furman, T. C.; Epp, J.; Hsiung, H. M.; Hoskins, J.; Long, G. L.; Mendelsohn, L. G.; Schoner, B.; Smith, D. P.; Smith, M. C. (1987) Recombinant Human Insulin-Like Growth Factor II Expressed in Escherichia coli. Nat. Biotechnol. 1987, 5, 1047−1051. (59) Lou, M.; Garrett, T. P.; McKern, N. M.; Hoyne, P. A.; Epa, V. C.; Bentley, J. D.; Lovrecz, G. O.; Cosgrove, L. J.; Frenkel, M. J.; Ward, C. W. The first three domains of the insulin receptor differ structurally from the insulin-like growth factor 1 receptor in the regions governing ligand specificity. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 12429− 12434.

1865

DOI: 10.1021/acs.accounts.7b00227 Acc. Chem. Res. 2017, 50, 1855−1865