Enhanced Synthetic Methods and Stabilization of Insulin-Like Peptides

Enhanced Synthetic Methods and Stabilization of Insulin-like Peptides

John Karas

ORCID ID: 0000-0002-3800-5152

PhD Thesis

February 2016

University of Melbourne, School of Chemistry

Florey Neurosciences Institute

CSIRO

Submitted in total fulfillment of the requirements of the degree of Doctor of Philosophy

Dedicated to Con and Sharon Karasmanis

Abstract

Basal-bolus insulin therapy is essential for maintaining tight glycemic control in patients with type 1 diabetes. Despite recent advances, there are issues with its long-term heat stability, which particularly affects patients living in poorer tropical regions. Hence there is a need to develop more thermally stable analogues of insulin. Some degradation products are caused by disulfide shuffling which leads to oligomerization. This could be avoided by substituting the disulfide bonds with non-reducible cystine isosteres such as cystathionine, but incorporation of non-natural mimetics in peptides generally requires chemical synthesis. All current insulin therapeutics on the market are produced biosynthetically due to the large mass of material required and there are currently no chemical methods available that rival these yields. A more efficient chemical synthesis is required to make a modified analogue viable.

Described herein are three key accomplishments which attempt to address the issues raised above. Firstly, an improved method for forming disulfide bonds in peptides utilizing a photocleavable thiol protecting group in combination with an “activated” thiol is reported. The efficacy of this new approach was established in simpler model systems and then applied to insulin, whereby higher yields were achieved compared to existing synthetic methods. Secondly, a proof-of- concept study was designed such that lactam bridges were incorporated into insulin-like peptide 3 as replacements for the intermolecular disulfide bonds. The insights gained from this study helped formulate new approaches for incorporating non-natural mimetics in insulin. Finally, replacement of the A6 – A11 disulfide bond with cystathionine in insulin was achieved via the use of orthogonally-protected monomeric building blocks. This new analogue was synthesized in excellent yield and has been fully characterized. It was found to have native binding activity for both insulin receptors and a tertiary structure similar to human insulin. Its serum stability was maintained and it exhibited improved thermal stability, which could lead to an enhanced therapeutic.

iii

Declaration

This is to certify that:

(1) The thesis comprises only my original work towards the PhD except where indicated in the Acknowledgments section. (2) Due acknowledgment has been made in the text to all other material used. (3) The thesis is fewer than 100,000 words in length.

Signed: ______

John Karas

Acknowledgments

Firstly, my sincerest thanks go to all those who assisted me in my work: Associate Professor Uta Wille who allowed me access to her photoreactor; Professor Richard Lewis and Dr Irina Vetter who performed the binding assays for α-conotoxin ImI; Associate Professor Briony Forbes who conducted all the insulin binding assays; Dr Fazel Shabanpoor who performed the RXFP2 binding assays for all INSL3 analogues; Dr Marco Sani who assisted me with CD spectroscopy, DSC and NMR the experiments.

I would also like to extend my gratitude towards my colleagues at the Bio21 Institute, who supported me in many different ways over the years. In particular I would like to thank Associate Professor Paul Donnelly, Associate Professor Craig Hutton, Dr Jade Cottam and Dr Veronica Borrett.

I would like to send my best wishes to all in the Wade and Hossain group at the Florey Neurosciences Institute: Assistant Professor Keiko Hojo, Dr Fazel Shabanpoor, Dr Julien Tailhades, Dr Vinojini Nair, Nitin Patil, Wenyi Li and Elaheh Jamasbi. It was a pleasure to work in this lab and I learned a great deal from you all. Most importantly, I will never forget the laughs we had and the beers we drank.

Special thanks go to Dr James Gardiner from the CSIRO. His help, both scientifically and financially, was greatly appreciated and he has been a good friend.

I would like to express my gratitude towards Dr Akhter Hossain, who was one of the main driving forces for allowing me the chance to do a PhD. I thank him for all his friendship and support over the last four years.

Thanks also to Professor John Wade who helped with my PhD candidature, provided funding and allowed me to follow my ideas through in the lab. He gave me many opportunities to speak at various symposia and offered excellent scientific and career advice.

I would like to extend my gratitude towards Professor Frances Separovic who has been an excellent mentor and provided me with a lot of support throughout my candidature. Her expertise in the biophysical experiments that were conducted was invaluable and I would like to thank her for that.

Special thanks go to Dr Denis Scanlon, who employed me at Auspep Pty Ltd all those years ago. He taught me everything he knew about peptides and gave me many opportunities over a long period of time. He has also been a great mentor and friend.

I would also like to express my gratitude to all my family and friends - you know who you all are! Thanks for putting up with me when I’ve been stressed or unavailable.

Thanks to my brother Andrew Karasmanis who has been going through a stressful time but has still managed to help out with family matters.

Finally, I would like to thank my parents Con and Sharon Karasmanis. They started with almost nothing in life but worked hard to provide me with the opportunity of having a great education, and always put my needs before theirs.

Table of Contents

Abbreviations ...... ix

Publications Derived from this Work ...... xiii

Oral Presentations Derived from this Work ...... xiii

Poster Presentations Derived from this Work ...... xiv

Chapter 1: Introduction ...... 1 1.1 Diabetes Mellitus ...... 2 1.2 Diabetes and Insulin ...... 3 1.3 Current Insulin Therapeutics ...... 6 1.4 Synthesis of Insulin ...... 12 1.5 Cystine Isosteres ...... 19 1.6 Aims of this Study ...... 21 1.7 References ...... 22

Chapter 2 ...... 35 2.1 Abstract ...... 36 2.2 Introduction ...... 36 2.3 Synthetic Methods ...... 38 2.4 Conclusion ...... 42 2.5 Acknowledgments ...... 44 2.6 References ...... 45

Chapter 3 ...... 51 3.1 Abstract ...... 52 3.2 Introduction ...... 52 3.3 Materials and Methods ...... 56 3.4 Results and Discussion ...... 58 3.5 Conclusion ...... 63 3.6 Acknowledgments ...... 63 3.7 References ...... 64

vii

Chapter 4 ...... 71 4.1 Introduction ...... 72 4.2 Synthesis of Fmoc-γ-Br-hAla-OH ...... 79 4.3 Synthesis of Mmt-Cys-OAll ...... 80 4.4 Synthesis of Human Insulin A-Chain ...... 81 4.5 Synthesis of B-Chain ...... 87 4.6 Combination of A-Chain and B-Chain ...... 90 4.7 Formation of the A7 – B7 Disulfide Bond...... 92 4.8 Conclusion ...... 94 4.9 References ...... 96

Chapter 5 ...... 103 5.1 Introduction ...... 104 5.2 Receptor Binding Activity...... 105 5.3 Circular Dichroism Spectroscopy ...... 107 5.4 1H Nuclear Resonance Spectroscopy ...... 108 5.5 Differential Scanning Calorimetry ...... 110 5.6 Human Serum Stability Assay ...... 111 5.7 Thermal Stability Assay ...... 113 5.8 Conclusions ...... 114 5.9 References ...... 116

Chapter 6 ...... 121 6.1 Introduction ...... 122 6.2 Synthesis of Insulin via 2-Nitroveratryl Thiol Protection ...... 122 6.3 Incorporation of Stable Cystine Isosteres in Insulin-like Peptides ...... 125 6.4 Summary ...... 131 6.5 References ...... 132

Appendix 1 ...... 134

Appendix 2 ...... 153

viii

List of Figures

Figure 1.1 ...... 4 Figure 1.2 ...... 4 Figure 1.3 ...... 6 Figure 1.4 ...... 10 Figure 1.5 ...... 14-15 Figure 1.6 ...... 18 Figure 1.7 ...... 18 Figure 2.1 ...... 38 Figure 2.2 ...... 38 Figure 2.3 ...... 40 Figure 2.4 ...... 41 Figure 2.5 ...... 43 Figure 3.1 ...... 55 Figure 3.2 ...... 59 Figure 3.3 ...... 59 Figure 3.4 ...... 61 Figure 3.5 ...... 61 Figure 3.6 ...... 62 Figure 4.1 ...... 72 Figure 4.2 ...... 75 Figure 4.3 ...... 77 Figure 4.4 ...... 80 Figure 4.5 ...... 81 Figure 4.6 ...... 82 Figure 4.7 ...... 85 Figure 4.8 ...... 86 Figure 4.9 ...... 89 Figure 4.10 ...... 89 Figure 4.11 ...... 89 Figure 4.12 ...... 91 Figure 4.13 ...... 92

Figure 4.14 ...... 94 Figure 4.15 ...... 94 Figure 5.1 ...... 106 Figure 5.2 ...... 107 Figure 5.3 ...... 109 Figure 5.4 ...... 110 Figure 5.5 ...... 112 Figure 5.6 ...... 114 Figure 6.1 ...... 124 Figure 6.2 ...... 129 Figure 6.3 ...... 130 Figure A1.1 ...... 139 Figure A1.2 ...... 140 Figure A1.3 ...... 141 Figure A1.4 ...... 143 Figure A1.5 ...... 143 Figure A1.6 ...... 144 Figure A1.7 ...... 144 Figure A1.8 ...... 146 Figure A1.9 ...... 146 Figure A1.10 ...... 147 Figure A1.11 ...... 147 Figure A1.12 ...... 149 Figure A1.13 ...... 150 Figure A1.14 ...... 151 Figure A2.1 ...... 157 Figure A2.2 ...... 158

Abbreviations

2-PhiPr 2-Phenylisopropyl Acm Acetamidomethyl AcOH Acetic acid Ala Alanine Alloc Allyloxycarbonyl API Active pharmaceutical ingredient Arg Arginine Asn Asparagine Asp Aspartic acid Boc tert.-Butyloxycarbonyl Bzl Benzyl cAMP Cyclic adenosine monophosphate CD Circular dichroism Cys Cysteine Dab Diaminobutyric acid DCM Dichloromethane Dde 1-(4,4-Dimethyl-2,6-dioxocyclohexylidene)-ethyl DIC Diisopropylcarbodiimide DIEA Diisopropylethylamine Dmab 4-{N-[1-(4,4-Dimethyl-2,6-dioxocyclohexylidene)-3- methylbutyl]-amino} benzyl DMAP Dimethylaminopyridine DMF N,N-Dimethylformamide DNA Deoxyribonucleic acid DODT 3,6-Dioxa-1,8-octane-dithiol DPDS Dipyridine disulfide Dpm Diphenylmethyl DPPA Diphenylphosphoryl azide Dpr Diaminopropionic acid DSC Differential scanning calorimetry E. coli Escherichia coli

ESI-MS Electrospray ionization Fm Fluorenylmethyl Fmoc 9-Fluorenylmethyloxycarbonyl Fmoc-OSu 9-Fluorenylmethyl N-succinimidyl carbonate Gln Glutamine Glu Glutamic acid Gly Glycine Gn.HCl Guanidimium hydrochloride HATU 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b] pyridinium 3-oxid hexafluorophosphate HBr Hydrobromic acid HEK-293T Human embryonic kidney 293 cells HEPES 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid HF Hydrogen fluoride His Histidine HIV Human immunodeficiency virus hNP2 human Neutrophal defensin peptide 2 HOBt 1-Hydroxybenzotriazole

IC50 Half maximal inhibitory concentration IGF-I Insulin-like growth factor I IGF-II Insulin-like growth factor II INSL3 Insulin-like peptide 3 IR-A Insulin receptor isoforms A IR-B Insulin receptor isoforms B ivDde 1-(4,4-Dimethyl-2,6-dioxocyclohexylidene)- 3-methylbutyl Lys Lysine MALDI Matrix-assisted laser desorption/ionization MeOH Methanol Met Methionine Mmt mono-Methoxytrityl Mmt-Cl mono-Methoxytrityl chloride MRE Mean residual weight ellipticity

xii

MS Mass spectrometry Mtt mono-Methyltrityl NaOH Sodium Hydroxide NCL Native chemical ligation NMP N-Methyl-2-pyrrolidone NMR Nuclear magnetic resonance NPH insulin Neutral Protamine Hagedorn insulin oNb ortho-Nitrobenzyl oNv ortho-Nitroveratryl Orn Ornithine PEG Polyethylene glycol PGA Penicillin G acylase PhAcm Phenylacetamidomethyl pI Isoelectric point Pro Proline PS Polystyrene Pyr Pyridinesulfenyl QTOF Quadrupole time-of-flight RP-HPLC Reversed phase high performance liquid chromatography RPMI medium Roswell Park Memorial Institute medium RXFP2 Relaxin/insulin-like family peptide receptor 2 Sec Selenocysteine SPPS Solid-phase peptide synthesis StBu Thio-tert.-butyl tBu tert.-Butyl TEA Triethylamine TFA Trifluoroacetic acid TFMSA Trifluoromethanesulfonic acid THF Tetrahydrofuran Thr Threonine Thz Thiazolidine TIPS Triisopropylsilane

xiii

Trp Tryptophan Trt Trityl Tyr Tyrosine α7 nAChR Alpha-7 nicotinic acetylcholine receptor

xiv

Publications Derived from this Work

Karas, J., Shabanpoor, F., Hossain, M. A., Gardiner, J., Separovic, F., Wade, J. D., Scanlon, D. B. "Total Chemical Synthesis of a Heterodimeric Interchain Bis- Lactam-Linked Peptide: Application to an Analogue of Human Insulin-Like Peptide 3." Int. J. Pept. 2013, vol. 2013, Article ID 504260, 8 pages.

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 (31), 9549–9552.

Oral Presentations Derived from this Work

“2-Nitroveratryl as a Novel Thiol Protecting Group for Directed Synthesis of Cysteine-rich Bioactive Peptides”, presented at the Neuropeptides division meeting, Florey Neurosciences Institute, 9th of October 2013.

“2-Nitroveratryl as a Novel Thiol Protecting Group for Directed Synthesis of Cysteine-rich Bioactive Peptides”, presented at the 4th Modern Solid Phase Peptide Synthesis & Its Applications Symposium”, Kobe Japan, 2nd of November 2013.

“2-Nitroveratryl as a Novel Thiol Protecting Group for Directed Synthesis of Cysteine-rich Bioactive Peptides”, presented at the Peptide Users Group Symposium, Florey Neurosciences Institute, 25th of November 2013

“2-Nitroveratryl as a Novel Thiol Protecting Group for Directed Synthesis of Cysteine-rich Bioactive Peptides”, presented at the CSIRO materials science and engineering program meeting, Clayton Victoria, 17th of December 2013.

Poster Presentations Derived from this Work

John A. Karas, Fazel Shabanpoor, Dick Wettenhall, John Wade and Denis Scanlon. “Lactam Bridges as Disulfide Mimics in INSL3: Solid Phase Synthesis Strategy”. Presented at the 6th International Conference on Relaxin and Related Peptides, Florence Italy, October 2012.

John A. Karas, Denis Scanlon, James Gardiner, Briony Forbes, Richard Lewis, Irina Vetter, Frances Separovic, Mohammed Akhter Hossain, John D. Wade. “2-Nitroveratryl as a Novel Thiol Protecting Group for Directed Synthesis of Cysteine-rich Peptides”. Presented at the 4th Asia-Pacific Peptide Synthesis Symposium, Osaka Japan, November 2013.

John A. Karas, Nitin Patil, Julien Tailhades, Marco Sani, Denis B. Scanlon, Briony Forbes, James Gardiner, Frances Separovic, John D. Wade, Mohammed Akhter Hossain. “Incorporation of Cystathionine as a Cystine Mimetic in Insulin to Develop More Stable Analogues”. Presented at the 11th Australian Peptide Conference, Kingscliff Australia, October 2015.

xvi

xvii

Chapter 1

Introduction

1.1 Diabetes Mellitus

Diabetes mellitus is a disease in which blood sugar levels are poorly regulated, leading to hyperglycemia and ketoacidosis. Common symptoms of this condition include polyphagia, polydipsia and polyuria. If left untreated then other symptoms such as weight loss, cognitive impairment, organ damage and coma can occur which eventually leads to death.1 There are three main forms of the disease: (1) type I or juvenile diabetes, (2) type II or adult onset diabetes, and (3) gestational diabetes.

Type I diabetes is an autoimmune disease which causes the destruction of the beta cells in the pancreas that produce insulin, a peptide hormone which regulates blood glucose levels by mediating uptake of glucose in to cells. The underlying cause is not clear although there is speculation that there are a combination of factors such as genetics and exogenous agents.2 The primary treatment for this form of the disease is therapy whereby insulin-based drugs are administered subcutaneously as a replacement for the endogenous hormone. Type II diabetes is more widespread and is caused by insulin resistance in which there is a lack of sensitivity of the insulin receptor towards its cognate ligand. It is closely linked with obesity which can be caused by deleterious lifestyle factors such as a lack of exercise, poor diet, stress and other psychological disorders. Genetic factors and other medical issues can also play a causal role in the onset of the condition.3 Typical treatments include a recalibration towards healthier lifestyle choices and the use of insulin secretagogues such as glucagon-like peptide-1 receptor agonists.3 Insulin-based therapeutics are also administered on occasions but are generally not the first treatment option. Gestational diabetes occurs during pregnancy and is due to insulin resistance because of dysfunctional insulin receptors. The condition is usually temporary although there is an increased risk for the mother and child to develop type II diabetes later in life. Diet and mild exercise are the primary treatments, although insulin therapeutics are used on occasions.4

It is estimated that there are 382 million people living with diabetes mellitus and a further 316 million high risk individuals with impaired glucose tolerance. It is

2 expected that worldwide, there will be 471 million with the disease by 2035 with about 80% living in low and middle income countries. In Australia alone, it is estimated that 10% of the adult population have some form of the disease.5 The global cost of diabetes is astronomical and strains many nations’ health budgets. It has been estimated that USD $548 billion is spent on the disease per annum, and this figure is expected to reach approximately $627 billion by 2035.5 Hence there is a pressing need to minimize these crippling costs borne by health budgets around the world. This can be done by addressing obesity and lifestyle factors but also by developing more effective therapeutics for the disease.

1.2 Diabetes and Insulin

As stated above, insulin is an essential therapeutic for the treatment of diabetes, particularly for type I. Given the worsening epidemic, it is anticipated that there will be a rising demand for insulin-based drugs in the near future, which will have a negative impact on government health budgets and put lives at risk. Insulin therapeutics are currently made recombinantly but continued research in to their synthesis is necessary since improved biosynthetic (and possibly chemical) yields could potentially reduce the cost of the medications. Further optimization of its pharmacological properties should also be pursued and an improved chemical synthesis can aid in this by allowing incorporation of unnatural mimetics into the hormone. These issues will be addressed in this work.

Insulin is a member of the insulin/relaxin super-family of peptide hormones, which also consists of insulin-like growth factors (IGF) I and II, plus seven members of the relaxin-like peptide family (Figure 1.1). The taxonomy of these peptides is characterized by a common ancestral past and their structural similarities.6 Homologs are also present in other species such as apes, mice, frogs and fish.7 The distinguishing structural features of this family of hormones include two inter-chain disulfide bonds connecting an A- and a B-chain as a

3 parallel heterodimer, plus a third intra-chain disulfide bond on the A-chain. This is illustrated in Figure 1.2 which shows the primary structure of human insulin.

The history of insulin can be traced back to the late 1800s. After a link was established between the pancreas and diabetes,8 it was soon discovered that destruction of the “islets of Langerhans” – the hormone-producing beta cells of the pancreas – led to elevated blood glucose levels and hence diabetes mellitus.9 In 1922, Banting and Best were able to isolate an extract of these islet cells exclusively from a canine pancreas. They then administered it intravenously to diabetic dogs and successfully lowered blood glucose levels in

4 the animals.10 Soon after, fourteen year-old diabetic Leonard Thompson was treated via subcutaneous injection with a bovine pancreatic extract, which was much easier to isolate. Using this treatment, the patient’s blood sugar concentration was lowered to normal levels, glycosuria was abolished, acetone disappeared from his urine, increased utilization of carbohydrates was observed and his general condition improved substantially.11 This was a remarkable success and the boy lived for thirteen more years. Soon after, the production of what then became known as insulin was refined, scaled up and sold by the pharmaceutical company Eli Lilly.

The actual structure of bovine insulin was not known until 1955 when Sanger, together with his co-workers, deduced the primary structure of each chain and the disulfide connectivity,12 winning him the 1958 Nobel Prize in chemistry. This led to attempts by three laboratories to chemically synthesize different forms of the hormone via a solution-phase approach,13–15 and also a solid-phase one.16 These syntheses confirmed the structure that was reported by Sanger. Unfortunately the yields were unacceptably low in all cases so these were not feasible methods to produce the hormone in large quantities. The three- dimensional structure of dimeric porcine insulin was elucidated soon after via X- ray crystallography by Hodgkin et al.17 There was much debate about how insulin is formed given its unusual hetero-dimeric nature. One possibility was that each chain was formed separately and then combined via disulfide bond formation. However, Steiner determined that a single chain pre-cursor is biosynthesized first as what is now known as proinsulin.18 This also contains a C-peptide with a B-C-A configuration that facilitates the folding of the pre- prohormone to the correct disulfide connectivity. It is then cleaved enzymatically to render the native hormone (Figure 1.3).

In order to minimize the use of animal glands, improve yields and avoid allergic reactions from non-human insulin therapeutics, there was a need to develop a reliable method for producing the hormone. In the 1970s, recombinant DNA technology was developed and the first biosynthesis of human insulin was subsequently reported by researchers at Genentech. Both chains were

5 expressed in Escherichia coli separately followed by their isolation, S-sulfonation and combination via oxidative disulfide bond formation.19 Subsequently developed methods express proinsulin analogues followed by folding and enzymatic cleavage of the C-peptide.20 All insulin therapeutics currently on the market are produced by the latter technique. There have been many attempts to synthesize insulin via chemical means (see Section 1.4). Although none of these methods have superseded the current biosynthetic route for existing insulin drugs, these chemical approaches are important because of the challenge afforded by such a complex yet small protein and also, in the future, they may allow the incorporation of non-natural mimetics. This could lead to insulin analogues with enhanced pharmacological properties.

1.3 Current Insulin Therapeutics

One of the key challenges for treating diabetes with insulin therapeutics is to be able to reproduce endogenous levels of the hormone. This is difficult since basal levels of insulin must be maintained as well as spikes in concentration, which is necessary during feeding. If there is no tight glycemic control in patients then there is an elevated risk of hypoglycemic episodes and organ damage over time. There are various insulin therapeutics on the market which

6 can help mitigate the risk of these symptoms. Some are long-acting in order to provide basal levels of the hormone. There are also short-acting insulin analogues which are administered as the bolus dosage in order to mimic endogenous prandial insulin levels. Most patients typically take a long-acting analogue and a short-acting one as part of their basal-bolus regimen. Insulin is typically administered via subcutaneous injection although other modes of delivery such as dermal, nasal, oral, buccal and pulmonary routes have been investigated.21

One of the characteristics of insulin at therapeutically useful concentrations is its propensity to form dimers and then hexamers when coordinated to zinc. These higher oligomers help stabilize insulin-based formulations and are more slowly absorbed in to the bloodstream. This is ideal for basal dosage delivery but makes it difficult to administer effective bolus doses. Therefore much research has been conducted in order to try to destabilize these oligomers. Three approaches were pursued: (1) repulsion of monomers via juxtaposed negative charges, (2) replacement of residues that participate in intermolecular hydrophobic interactions with hydrophilic residues, and (3) replacement of the histidine, which is the metal binding site.22 Insulin Lispro was the first fast-acting insulin drug. Proline at position B28 was previously identified as being critical for high-affinity self-assembly. In order to disrupt this binding, lysine at position B29 was swapped with the proline residue (Figure 1.4(i)). This diminished the stability of the oligomeric form, allowing faster dissociation in the subcutaneous depot and hence more rapid absorption in to the circulatory system, resulting in an onset of action of 10 – 15 minutes. Importantly, though, hexameric insulin could still be formed which is desirable given its stabilizing effect.23 Insulin Aspart uses a similar strategy whereby proline at B28 is substituted with an aspartate residue, which is anionic at physiological pH. This disrupts the monomer-to-monomer interaction with glycine B23, and there is also a further repulsion with a glutamate residue at B21 (Figure 1.4(ii)).24 The development of insulin Glulisine used similar logic and contains two substitutions on the B- chain: asparagine substituted with lysine at B3, and lysine substituted with the charged glutamate at position B29 (Figure 1.4(iii)).25 All three of these fast-

7 acting analogues more closely mimic post-prandial levels of insulin when administered.

The first longer-acting drug that came on to the market was Neutral Protamine Hagedorn (NPH) insulin. This was formulated as a cloudy mixture of protamine which slows absorption and a hexameric zinc complex of porcine insulin. Its duration of action is approximately 15-18 hours; however, there is an inherent problem with this medication since insulin levels peak after six hours. Therefore it cannot be taken before bedtime as it could lead to nocturnal hypoglycemia. A recombinant human version of the formulation has also been developed but this exhibits high variability with respect to dosage due to inconsistent sampling from patients and uneven absorption after subcutaneous injection because it is a precipitate.26 In order to avoid the injection of suspensions and thus improve the consistency of the dose response, insulin Glargine was developed. Asparagine at A20 was replaced with glycine and two arginine residues were added to the C-terminus of the B-chain (Figure 1.4(iv)). This analogue is formulated as a clear solution in acidic media. Its isoelectric point (pI) is shifted towards neutral pH such that the analogue is insoluble after injection due to the physiological environment. This slows the absorption of the drug in to the capillaries and results in a flatter pharmacokinetic profile than NPH insulin and a lower risk of nocturnal hypoglycemia. However because of its acidic formulation, insulin glargine is unable to be mixed with other short-acting therapeutics and can possibly lead to pain during injection.27 Furthermore, it has a higher affinity towards the insulin-like growth factor-1 (IGF-I) receptor than native insulin which leads to concerns that it may be more mitogenic.28 Insulin Detemir has a similar pharmacokinetic profile to glargine but can be formulated as a pH neutral solution. The C-terminus of the B-chain is truncated, eliminating the threonine residue at B30 and the lysine at B29 is acylated with a myristyl moiety (Figure 1.4(v)). The fatty acid stabilizes the insulin hexamer which slows absorption into the capillaries after injection and also binds to albumin which can further delay its action in plasma.29 Insulin Degludec is a further improvement for delivery of basal levels of insulin. Similar in structure to insulin Detemir, it contains a fatty di-acid instead of a myristyl group (Figure 1.4(vi)). It lasts for more than 24

8 hours as it can form multi-hexameric oligomers which enhances its stability.30 It is also soluble in vivo which means a slow, consistent absorption of insulin occurs, resulting in a very flat pharmacokinetic profile.31 Short-acting insulins can be added to the formulation which means that they can be administered as a single basal-bolus therapy. These significant advances in glycemic control for diabetics have allowed patients to lead relatively normal lives. Furthermore, the most recent therapeutics with optimized pharmacokinetic profiles should minimize complications associated with the disease and lead to longer lifespans.

But there are still a number of challenges remaining regarding treatment of diabetes. Insulin formulations must be refrigerated otherwise degradation will occur particularly when the hexamer disassociates. This is especially the case for short-acting analogues where the protective oligomeric zinc complex has been deliberately destabilized by point mutations. Brange and co-workers determined that there are a number of degradation products which can arise in pharmaceutical preparations. These include trans-amidation of the A21 asparaginyl residue in acidic preparations and deamidation at B3 in neutral ones; the latter can also form an iso-aspartyl residue. Hydrolytic cleavage of the A8 – A9 peptide bond can occur in certain preparations.32 Covalent dimers and higher oligomers can also be formed via acylation of the B-chain N-terminus (predominantly) with side-chain carboxyamide groups,33 or cleavage of the A7 – B7 disulfide bond and subsequent intermolecular thiolysis.34 Head-to-tail insulin fibrillation can also occur via exposure of hydrophobic residues of a partially unfolded B-chain C-terminal segment35 which is a similar mechanism to its binding to the insulin receptor.36 Over time, significant levels of these modifications can affect dosage and cause immunological side effects.34 These degradative processes in insulin formulations are exacerbated by at least a factor of 10 for each 10 oC increment.37 Approximately 80% of diabetes cases occur in the third world or developing countries,5 many of whom live in tropical climates and lack consistent access to electricity and refrigeration. If the insulin is not stored properly then this can complicate treatment and lead to poor health outcomes.37 Therefore there is a need to develop more heat-stable analogues

10 in order to ensure that the disease can be effectively treated in all parts of the world.

There are a number of strategies that can be employed in order to minimize degradation of insulin in pharmaceutical preparations. Deamidation can be avoided by substitution of the asparagine residues. Point mutations in insulin Glargine (asparagine substituted with glycine at A21) and insulin Glulisine (asparagine substituted with lysine at B3) are examples of this, thus enhancing the stability of the active pharmaceutical ingredient (API). Fibrillation of insulin is also problematic. Hubálek and co-workers have recently reported an insulin analogue with four disulfide bonds which does not aggregate.38 This also exhibits enhanced thermodynamic stability as does a single chain analogue with a shortened C-peptide.37 As discussed above, the stability of the disulfide bonds is also an issue, particularly the A7 – B7 cystine bridge. Therefore in order to block cleavage of these sensitive disulfide bonds, non-reducible cystine isosteres could potentially be incorporated as their replacement. The efficacy of this tactic has been demonstrated by substitution of cystine with selenoether bridges in oxytocin. It was found that there was over a twenty-fold increase in the half-life for one analogue with respect to thermal stability (conducted at 55 oC). This modification also achieved a three- to four-fold increase in stability when incubated in human serum.39

Incorporation of a non-reducible construct avoids the reduction and concomitant scrambling of disulfides which is caused by redox enzymes and glutathione.40 Furthermore, efficient incorporation of these mimetics (in combination with other stabilizing strategies) could lead to the development of orally available analogues. Although ambitious, delivery via this route would lead to increased patient compliance and greater ease of storage. Efforts are already underway to achieve this goal41 but is beyond the scope of this study. But one of the challenges of incorporating non-proteinogenic constructs in peptides is that chemical synthesis is usually required. This is difficult given that the strategy must be feasible for scale-up and that native human insulin is also complex in structure with 51 residues, two chains and three disulfide bonds. However,

11 there are precedents for long peptide active pharmaceutical ingredients (APIs) that are made chemically such as the 36 residue Fuzeon HIV fusion inhibitor and 41 amino acid corticotropin-releasing factor.42 In the case of insulin, development of new, higher-yielding synthetic methodologies should be pursued before attempts are made to incorporate stable cystine isosteres.

1.4 Synthesis of Insulin

Large-scale synthesis of insulin is achieved biosynthetically. This is accomplished by either expression of both chains separately followed by combination or expression of proinsulin and subsequent enzymatic cleavage of the C peptide.43 The latter is the current preferred method. But as stated above, this limits the scope for incorporation of non-natural structures such as cystine mimetics. Hence there has been significant effort towards developing a more robust, cost-effective and high-yielding chemical route that is scalable. This will make the development of insulin peptidomimetics with enhanced pharmacokinetic properties feasible as therapeutics.

There are many examples in the literature reporting on the chemical synthesis of insulin analogues. Bovine insulin was first chemically synthesized via classical solution-phase techniques independently by three laboratories in the 1960s.13–15 The first solid-phase synthesis of insulin (bovine) was reported by Marglin and Merrifield in 1966 using a tert.-butyloxycarbonyl/benzyl (Boc/Bzl) strategy.16 The speed of assembly for both chains via this method was an improvement on previous efforts. For all of these methods, the free sulfhydryls on both chains were converted to their S-sulfonate form in order to enhance their solubility in aqueous media and hence their ease of purification. All cysteine residues were then reduced back to thiols and, simultaneously, both chains were combined and folded in a one-pot fashion (Figure 1.5(i)). This produced the target peptide plus mixed disulfide products and homo-dimers.

Several methods have been developed in order to form each disulfide bond regioselectively via the use of orthogonal S-protecting groups. Sieber, Rittel and

12 co-workers reported the first chemical synthesis of insulin using this approach in 1977 by assembling a number of semi-protected fragments together in solution. In the final step, the N-terminal intermolecular disulfide bond was formed via oxidative cleavage of acetamidomethyl protected A7 and B7 cysteine residues (Figure 1.5(ii)).44 Although the yield for the final oxidation step was ~70%, the overall yield of synthetic insulin was ~1%. In 1993, Kiso and co-workers developed a 9-fluorenylmethyloxycarbonyl (Fmoc) solid-phase protocol, using a combination of S-trityl (Trt), S-acetamidomethyl (S-Acm) and S-tert.-butyl (tBu) protection (Figure 1.5(iii)). After dimerization of the A and B-chains via S- pyridinesulfenyl (Pyr) mediated ligation, the final two cystine bonds were formed via an iodine oxidation followed by treatment with a silyl chloride.45 The overall yield for this synthesis was still quite low at ~1%. Wade and co-workers developed a similar method for the synthesis of insulin-like peptides, but formed the intramolecular A-chain first, followed by S-Pyr functionalization, dimerization and finally the iodine oxidation (Figure 1.5(iv)).46–48 This protocol has been employed widely for other peptides because it avoids the use of a silyl chloride which can chlorinate tryptophan. Recently, Mezo et al incorporated iso-acyl dipeptides which introduce temporary amino groups to both chains. These surrogates enhanced the solubility at acidic pH levels which enabled an easier purification and led to higher yields (Figure 1.5(v)).49 Regioselective methods have become very important for structure-function studies since non-directed folding can be inconsistent when synthesizing mutations of the parent peptide.

Alternative approaches have been pursued, largely based on single-chain precursors with shortened C-peptide surrogates. Shin and co-workers incorporated a nonapeptide with a reverse turn in order to enhance the yields of the refolding step (a 20-40% improvement compared to native pro-insulin).50 Hoeg-Jensen et al minimized the length of the C-peptide to 5 residues and inserted glutamate residues in order to enhance its solubility at high pH, achieving a folding yield of ~25% (Figure 1.5(vi)).51 There have also been reports of a single residue methioninyl C-peptide.52 Kent and co-workers have

15 also published a number of novel methods: ligation and disulfide folding via formation of an ester bond between the A4 glutamate and the B30 threonine side-chains (Figure 1.5(vii)),53 and using an oxime-forming ligation to tether both chains through the A-chain N-terminus and the lysine side-chain at position B29. This surrogate C-peptide is PEGylated to improve solubility and is enzyme-cleavable (Figure 1.5(viii)).54 Nevertheless, there is still much scope for further improving the synthetic yields of insulin. It was thought that pursuing a regioselective strategy was appropriate as it would ensure that the remaining two disulfide bonds have the correct connectivity. An alternative method to the final iodine oxidation would significantly improve yields and was pursued. In general, mass recoveries for this step are widely variable and in the order of 10- 20%, often due to disulfide shuffling, modification of methionine, tryptophan and, in the case of insulin, tyrosine residues.

A number of different S-protecting groups have been developed which could be suitable for replacing S-Acm protection, given its limitations. These must be orthogonal (or at least semi-orthogonal) to the acid labile S-Trt, S-tBu and S-Pyr systems. Furthermore, cystine bonds must be stable to the cleavage conditions, and the S-protecting group should be stable to assembly via Fmoc SPPS. A thiol protecting group based on the S-phenylacetamidomethyl (S-PhAcm) moiety which is enzymatically cleaved via penicillin G acylase from E. coli (PGA), has been developed and applied to the assembly of disulfide rich peptides.55 Importantly, the stability of S-PhAcm is analogous to S-Acm, which means that it could potentially be employed in the synthesis of insulin. Further development of this work has led to PGA being immobilised onto a solid support allowing the deprotection of S-PhAcm and convenient removal of the enzyme via a simple filtration step.56 A second possibility is the use of hydrazine-labile S-protection which has been recently developed by Liu et al.57 It is also based on an acylamidomethyl-type structure and possesses a bicyclic ring system. Human neutrophal defensin (hNP2), a 29 mer with three disulfide bonds was used as a model system to demonstrate the utility of this approach. In this study, photocleavable protecting groups were investigated since they can be

16 cleaved under ambient conditions and often without the requirement of any other reagents.

Photocleavable protecting groups, such as those based on ortho-nitrobenzyl, benzoin, phenacyl, coumaryl and arylmethyl moieties, have been in use since the 1960s and are commonly used as cages (or photo-switches) for biomolecules.58 The ortho-nitrobenzyl based cages have been widely employed, and there are instances of them being used to block sulfydryls, which is relevant to this work. Bayley and co-workers used a hydrophilic α-carboxy-2-nitrobenzyl system to cage the free thiol in glutathione in solution via the bromide derivative followed by photocleavage at 300 nm as a proof of concept.59 Further work by this group used 2-nitrobenzyl bromide and 2-nitroveratryl bromide to label a short cysteine-containing synthetic peptide and were able to successfully uncage it.60 Lester et al succeeded in incorporating 2-nitrobenzyl-protected cysteine and tyrosine within a transmembrane segment of the nicotinic acetylcholine receptor via biosynthesis in order to investigate ion channels.61 A 2-nitrobenzyl cysteine building block with Nα Fmoc protection has also been synthesized for the purpose of incorporation into a peptide sequence to facilitate a deprotection-conjugation protocol.62 Brik et al were able to show that Fmoc- Cys(2-nitrobenzyl)-OH was indeed compatible with Fmoc-SPPS protocols whereby they assembled and isolated a 31 mer in good yield.63 Further, introduction of two alternative photocleavable S-protecting groups into a bis-cysteinyl peptide was achieved. The aim of this work was to demonstrate the viability of wavelength-selective uncaging which was achieved via irradiation at 430 nm to cleave an S-coumaryl-based protecting group followed by 325 nm to cleave an S-2-nitrobenzyl based one.64 These examples highlight the amazing versatility of photochemistry and its potential for orthogonally cleaving S-protecting groups during regioselective disulfide bond formation.

It was decided that the 2-nitroveratryl (4,5-dimethoxy-2-nitrobenzyl, oNv) moiety should evaluated first as protection for the β-thiol of cysteine. 2-nitrobenzyl- based cages do have disadvantages, one being that the 2-nitrosoarylaldehyde photocleavage product is potentially reactive. These systems also suffer from

17 slower reaction kinetics compared with other photo-caging systems; this is exacerbated by the absorbance of the nitroso side product. However, these factors will most likely have a minor impact because the working concentrations required for this work will be highly dilute. It should also be noted that nitrobenzyl photocages have already been successfully used in peptides and Fmoc-SPPS. 2-nitroveratryl was preferred over the simpler 2-nitrobenzyl system because although the latter has a much higher quantum yield, the former has a stronger absorbance in the region of 350 – 400 nm. This is desirable since irradiation at shorter wavelengths can degrade sensitive residues such as tryptophan and tyrosine.58 The proposed mechanism for the photolysis reaction

18 is outlined in Figure 1.6, whereby cleavage of the protecting group occurs via a bicyclic intermediate.65 There are two possible ways to employ this chemistry for the formation of disulfide bonds. A symmetrical approach whereby both thiols are S-oNv protected and irradiated at 360 nm in the presence of an oxidant is one method. A second option is an asymmetrical approach whereby one thiol is S-oNv protected and the other is S-Pyr protected (Figure 1.7). Although the latter method will require an extra synthetic step (S-Pyr derivatization), it is still preferred since the photocleavage and concomitant thiolysis will most likely lead to a minimum of mixed disulfide products. This latter method could be applied to insulin in order to form the final disulfide bond in greater yield than current tactics allow.

1.5 Cystine Isosteres

As stated earlier, the disulfide bonds in insulin are susceptible to disulfide shuffling during long storage in solution, particularly A7 – B7 cystine which leads to the formation of covalent dimers.34 A cystine mimic can be inserted in to the insulin structure in order to stabilize the inter-chain connectivity. There are numerous options that can be pursued in order to achieve this. Diselenides have been used widely since they closely mimic disulfides and are much more stable.66 This moiety has been incorporated in to a sunflower trypsin inhibitor which retained high potency67 and also an α-conotoxin which maintained full biological activity and had enhanced stability under biologically reducing conditions.66 More recently, a selenoether bond has been incorporated in to oxytocin.39 1,4 triazoles68,69 and 1,5-triazoles70 can also be useful cystine isosteres. Recently, the A7 – B7 disulfide bond in insulin was replaced with two triazole linkages, however both analogues were inactive.71 Numerous dicarba peptides have also been prepared and shown to possess near-native structure and extended in vivo stability.72–77 Mono-substituted dicarba analogues of the heterodimeric peptides relaxin-378 and insulin-like peptide 3 (INSL3)79 have also been prepared and evaluated. An A6 – A11 insulin dicarba analogue has also been patented.80 The incorporation of lactam bridges in to peptides as disulfide

19 mimetics has been widely reported in the literature.81–88 Typically, an aspartyl/2,3-diaminopropionyl pairing is used since this is isosterically similar to a cystine bond. Recently, Büllesbach and Schwabe prepared two lactam analogues of human INSL3 that replaced the two interchain disulfide bonds in order to study the role of the A11 – B10 and A24 – B22 cystine bonds in both binding and receptor activation.89 Cystathionine (or thioether) bridges have also been incorporated in to disulfide rich peptides, and it has been possible to incorporate not only one bridge90–95 but also two.77,96,97 Jošt and Rudinger have also synthesized insulin with an A6 – A11 cystathionine bridge, but in low yield.98

When deciding which cystine isosteres to use, thought was given to toxicity, geometry, ease of incorporation and potential yields given that the chemistry must be scalable. Although diselenides and selenoethers are excellent disulfide mimics, there may be safety concerns over the regular and long term use of selenocysteine-containing peptides. Basal-bolus insulin therapy requires daily injections over a lifetime and elevated levels of selenium may prove to be toxic to humans.99 Triazoles have a slightly different geometry and restricted rotation so there is potential for retardation of the three-dimensional structure which could render the peptide inactive.71 The toxicity of the triazole moiety is also unknown. Dicarba peptide analogues hold much promise; however, these are difficult to prepare and the peptide assembly and building block synthesis can be low-yielding. Although lactam bridges are not structurally or electronically similar to disulfide bonds, they do allow an appropriate geometry and are easy to synthesize as many of the required orthogonally protected building blocks are commercially available. However, cystathionine is the preferred isostere because it is an endogenous substance in humans (and therefore less likely to be toxic) and also very closely mimics the disulfide geometry. But there have been mixed results with respect to yields for incorporating this moiety in peptides, so more efficient chemical methods were developed. Since native insulin is particularly difficult to prepare, it was decided that a proof of concept study be undertaken first. INSL3, which is structurally similar to insulin, was selected as a model system because it is more soluble in aqueous media and

20 therefore easier to fold and purify. Furthermore, lactam bonds were incorporated as intermolecular bridges first since these analogues could be more rapidly prepared as there was no need to synthesize building blocks. The insights gained from this initial study helped determine the preferred synthetic approach for the insertion of cystathionine bridges in insulin. Although the A7 – B7 cystine bridge is apparently more unstable, incorporation of the A6 – A11 cystathionine bridge was performed since it is the most synthetically accessible analogue.

1.6 Aims of this Study

A number of challenges were overcome during the optimization of the synthesis of insulin and assembly of an analogue with improved thermal stability, and was based on the following goals:

1. Develop an enhanced chemical synthesis of human insulin by employing a photocleavable thiol protecting group that could be cleaved under ambient conditions, thus avoiding a low-yielding iodine oxidation (Chapter 2).

2. Undertake a proof of concept study to evaluate the feasibility of incorporating stable lactam bridges in INSL3 (Chapter 3).

3. Synthesize and characterize orthogonally-protected building blocks suitable for the generation of thioether bridges in peptides, and use them to assemble cystathionine analogues of insulin via new synthetic protocols (Chapter 4).

4. Characterize any synthetic cystathionine-containing insulin analogues in the following experiments: (1) insulin receptor binding assays, (2) circular dichroism spectroscopy, (3) 1H NMR spectroscopy, (4) differential scanning calorimetry, (5) human serum stability assay and (6) thermal stability assay (Chapter 5).

Success was dependent on creative method development and careful optimization of any new protocols that were established, as well as thorough biological and physico-chemical evaluation of new compounds.

1.7 References

(1) Chiang, J. L., Kirkman, M. S., Laffel, L. M. B., Peters, A. L. "Type 1 Diabetes Through the Life Span: A Position Statement of the American Diabetes Association." Diabetes Care 2014, 37 (7), 2034–2054.

(2) Knip, M., Veijola, R., Virtanen, S. M., Hyöty, H., Vaarala, O., Åkerblom, H. K. "Environmental Triggers and Determinants of Type 1 Diabetes." Diabetes 2005, 54 (suppl 2), S125–S136.

(3) Ripsin, C. M., Kang, H., Urban, R. J. "Management of Blood Glucose in Type 2 Diabetes Mellitus." Am Fam Physician 2009, 79 (1), 29–36.

(4) Donovan, P. J., McIntyre, H. D. "Drugs for Gestational Diabetes." Aust Prescr 2010, 33, 141–144.

(5) IDF Diabetes Atlas, Sixth Edition, International Diabetes Federation, 2013.

(6) Shabanpoor, F., Separovic, F., Wade, J. D. "Chapter 1: The Human Insulin Superfamily of Polypeptide Hormones." In Vitamins & Hormones, Elsevier, 2009, Vol. 80, pp 1–31.

(7) Wilkinson, T. N., Speed, T. P., Tregear, G. W., Bathgate, R. A. "Evolution of the Relaxin-like Peptide Family." BMC Evol. Biol. 2005, 5, 14.

(8) von Mering, J., Minkowski, O. "Diabetes Mellitus Nach Pankreas Extirpation." Arch. Für Exp. Pathol. Pharmakol. 1889, 26, 371–387.

(9) Opie, E. L. "Diabetes Mellitus Associated with Hyaline Degeneration of the Islands of Langerhans of the Pancreas." Bull. Johns Hopkins Hosp. 1901, 12, 263–264.

(10) Banting, F. G., Best, B. A. "The Internal Secretion of the Pancreas." J. Lab. Clin. Med. 1922, 7 (5), 251–266.

(11) Banting, F. G., Best, C. H., Collip, J. B., Campbell, W. R., Fletcher, A. A. "Pancreatic Extracts in the Treatment of Diabetes Mellitus." Can. Med. Assoc. J. 1922, 12 (3), 141.

(12) Ryle, A. P., Sanger, F., Smith, L. F., Kitai, R. "The Disulphide Bonds of Insulin." Biochem. J. 1955, 60 (4), 541.

(13) Meienhofer, V. J., Schnabel, E., Bremer, H., Brinkhoff, O., Zabel, R., Sroka, W., Klostermeyer, H., Brandenburg, D., Okuda, T., Zahn, H. "Synthese Der Insulinketten Und Ihre Kombination Zu Insulinaktiven Präparaten." Z. Für Naturforschung 1963, 18 b, 1120–1121.

(14) Katsoyannis, P. G., Fukuda, K., Tometsko, K., Suzuki, K., Tilak, M. Insulin "Peptide. X. The Synthesis of the B-Chain of Insulin and Its Combination with Natural or Synthetic A-Chain to Generate Insulin Activity." J. Am. Chem. Soc. 1964, 86, 930–932.

(15) Du, Y. C., Jiang, R. Q., Tsou, C. L. "Conditions for Successful Resynthesis of Insulin from Its Glycyl and Phenylalanyl Chains." Sci. Sin. 1965, 14, 229–236.

(16) Marglin, B., Merrifield, R. B. "The Synthesis of Bovine Insulin by the Solid Phase Method." J. Am. Chem. Soc. 1966, 88 (21), 5051–5052.

(17) Adams, M. J., Blundell, T. L., Dodson, E. J., Dodson, G. G., Vijayan, M., Maker, E. N., Harding, M. M., Hodgkon, D. C., Rimmer, B., Sheat, S. "Structure of Rhombohedral 2 Zinc Insulin Crystals." Nature 1969, 224 (5218), 491–495.

(18) Steiner, D. F. "Proinsulin and the Biosynthesis of Insulin." N Engl J Med 1969, 280, 1106–1113.

(19) Goeddel, D. V., Kleid, D. G., Bolivar, F., Heyneker, H. L., Yansura, D. G., Crea, R., Hirose, T., Kraszewski, A., Itakura, K., Riggs, A. D. "Expression in Escherichia Coli of Chemically Synthesized Genes for Human Insulin." Proc. Natl. Acad. Sci. 1979, 76 (1), 106–110.

(20) Ladisch, M. R., Kohlmann, K. L. "Recombinant Human Insulin." Biotechnol. Prog. 1992, 8 (6), 469–478.

(21) Cefalu, W. T. "Concept, Strategies, and Feasibility of Noninvasive Insulin Delivery." Diabetes Care 2004, 27 (1), 239–246.

(22) Brange, J., Vølund, A. "Insulin Analogs with Improved Pharmacokinetic Profiles." Adv. Drug Deliv. Rev. 1999, 35 (2), 307–335.

(23) Brems, D. N., Alter, L. A., Beckage, M. J., Chance, R. E., DiMarchi, R. D., Green, L. K., Long, H. B., Pekar, A. H., Shields, J. E., Frank, B. H. "Altering the Association Properties of Insulin by Amino Acid Replacement." Protein Eng. 1992, 5 (6), 527–533.

(24) Setter, S. M., Corbett, C. F., Campbell, R. K., White, J. R. "Insulin Aspart: A New Rapid-Acting Insulin Analog." Ann. Pharmacother. 2000, 34 (12), 1423–1431.

(25) Garg, S. K., Ellis, S. L., Ulrich, H. "Insulin Glulisine: A New Rapid-Acting Insulin Analogue for the Treatment of Diabetes." Expert Opin. Pharmacother. 2005, 6 (4), 643–651.

(26) Yki-Järvinen, H. "Combination Therapy with Insulin and Oral Agents: Optimizing Glycemic Control in Patients with Type 2 Diabetes Mellitus." Diabetes Metab. Res. Rev. 2002, 18 (S3), S77–S81.

(27) Vague, P., Selam, J. L., Skeie, S., De Leeuw, I., Elte, J. W., Haahr, H., Kristensen, A., Draeger, E. "Insulin Detemir Is Associated with More Predictable Glycemic Control and Reduced Risk of Hypoglycemia than NPH Insulin in Patients with Type 1 Diabetes on a Basal-Bolus Regimen with Premeal Insulin Aspart." Diabetes Care 2003, 26 (3), 590–596.

(28) Kurtzhals, P., Schäffer, L., Sørensen, A., Kristensen, C., Jonassen, I., Schmid, C., Trüb, T. "Correlations of Receptor Binding and Metabolic and Mitogenic Potencies of Insulin Analogs Designed for Clinical Use." Diabetes 2000, 49 (6), 999–1005.

(29) Kurtzhals, P. "Engineering Predictability and Protraction in a Basal Insulin Analogue: The Pharmacology of Insulin Detemir." Int. J. Obes. 2004, 28, S23–S28.

(30) Kalra, S. "Insulin Degludec: A Significant Advancement in Ultralong-Acting Basal Insulin." Diabetes Ther. 2013, 4 (2), 167–173.

(31) Heise, T., Hermanski, L., Nosek, L., Feldman, A., Rasmussen, S., Haahr, H. "Insulin Degludec: Four Times Lower Pharmacodynamic Variability than Insulin Glargine under Steady-State Conditions in Type 1 Diabetes." Diabetes Obes. Metab. 2012, 14 (9), 859–864.

(32) Brange, J., Langkjær, L., Havelund, S., Vølund, A. "Chemical Stability of Insulin. 1. Hydrolytic Degradation During Storage of Pharmaceutical Preparations." Pharm. Res. 1992, 9 (6), 715–726.

(33) Weiss, M. "Design of Ultra-Stable Insulin Analogues for the Developing World." J. Health Spec. 2013, 1 (2), 59.

(34) Brange, J., Havelund, S., Hougaard, P. "Chemical Stability of Insulin. 2. Formation of Higher Molecular Weight Transformation Products During Storage of Pharmaceutical Preparations." Pharm. Res. 1992, 9 (6), 727– 734.

(35) Brange, J., Andersen, L., Laursen, E. D., Meyn, G., Rasmussen, E. "Towards Understanding Insulin Fibrillation." J. Pharm. Sci. 1997, 86 (5), 517–525.

(36) Menting, J. G., Yang, Y., Chan, S. J., Phillips, N. B., Smith, B. J., Whittaker, J., Wickramasinghe, N. P., Whittaker, L. J., Pandyarajan, V., Wan, Z., Yadav, S. P., Carroll, J. M., Strokes, N., Roberts Jr, C. T., Ismail- Beigi, F., Milewski, W., Steiner, D. F., Chauhan, V. S., Ward, C. W., Weiss, M. A., Lawrence, M. C. "Protective Hinge in Insulin Opens to Enable Its Receptor Engagement." Proc. Natl. Acad. Sci. 2014, 111 (33), E3395–E3404.

(37) Hua, Q., Nakagawa, S. H., Jia, W., Huang, K., Phillips, N. B., Hu, S., Weiss, M. A. "Design of an Active Ultrastable Single-Chain Insulin Analog: Synthesis, Structure, and Therapeutic Implications." J. Biol. Chem. 2008, 283 (21), 14703–14716.

(38) Vinther, T. N., Norrman, M., Ribel, U., Huus, K., Schlein, M., Steensgaard, D. B., Pedersen, T. Å., Pettersson, I., Ludvigsen, S., Kjeldsen, T., Jensen, K. J., Hubálek, F. "Insulin Analog with Additional Disulfide Bond Has Increased Stability and Preserved Activity." Protein Sci. 2013, 22 (3), 296–305.

(39) Dantas de Araujo, A., Mobli, M., Castro, J., Harrington, A. M., Vetter, I., Dekan, Z., Muttenhaler, M., Wan, J., Lewis, R. J., King, G. F., Brierley, S. M., Alewood, P. F. "Selenoether Oxytocin Analogues Have Analgesic Properties in a Mouse Model of Chronic Abdominal Pain." Nat. Commun. 2014, 5, 3165.

(40) Gilbert, H. F. "Redox Control of Enzyme Activities by Thiol/disulfide Exchange." Methods Enzymol. 1984, 107, 330–351.

(41) Hamman, J. H., Enslin, G. M., Kotzé, A. F. "Oral Delivery of Peptide Drugs." BioDrugs 2005, 19 (3), 165–177.

(42) Bray, B. L. "Large-Scale Manufacture of Peptide Therapeutics by Chemical Synthesis." Nat. Rev. Drug Discov. 2003, 2 (7), 587–593.

(43) Kroeff, E. P., Owens, R. A., Campbell, E. L., Johnson, R. D., Marks, H. I. "Production Scale Purification of Biosynthetic Human Insulin by Reversed- Phase High-Performance Liquid Chromatography." J. Chromatogr. 1989, 461, 45–61.

(44) Sieber, P., Kamber, B., Hartmann, A., Jöhl, A., Riniker, B., Rittel, W. "Totalsynthese von Humaninsulin. IV. Beschreibung Der Endstufen." Helv. Chim. Acta 1977, 60 (1), 27–37.

(45) 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 (24), 11384–11392.

(46) Lin, F., Otvos, L., Kumagai, J., Tregear, G. W., Bathgate, R. A. D., Wade, J. D. "Synthetic Human Insulin 4 Does Not Activate the G-Protein- Coupled Receptors LGR7 or LGR8." J. Pept. Sci. 2004, 10 (5), 257–264.

(47) 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., Gundlach, A. L., Samuel, C. S., Tregear, G. W., Wade, J. D. "Relaxin-3: Improved Synthesis Strategy and Demonstration of Its High-Affinity Interaction with the Relaxin Receptor LGR7 Both In Vitro and In Vivo." Biochemistry (Mosc.) 2006, 45 (3), 1043–1053.

(48) 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 (3), 211–215.

(49) Liu, F., Luo, E. Y., Flora, D. B., Mezo, A. R. "A Synthetic Route to Human Insulin Using Isoacyl Peptides." Angew. Chem. Int. Ed Engl. 2014, 53 (15), 3983–3987.

(50) Chang, S., Kim, D., Choi, K., Shin, J., Shin, H. "Human Insulin Production from a Novel Mini-Proinsulin Which Has High Receptor-Binding Activity." Biochem J 1998, 329, 631–635.

(51) Tofteng, A. P., Jensen, K. J., Schäffer, L., Hoeg-Jensen, T. "Total Synthesis of desB30 Insulin Analogues by Biomimetic Folding of Single- Chain Precursors." ChemBioChem 2008, 9 (18), 2989–2996.

(52) Moroder, L., Buchner, J. Oxidative Folding of Peptides and Proteins, RSC biomolecular sciences, Royal Society of Chemistry, 2009.

(53) Sohma, Y., Hua, Q., Whittaker, J., Weiss, M. A., Kent, S. B. H. "Design and Folding of [GluA4(OβThrB30)]Insulin (“Ester Insulin”), a Minimal Proinsulin Surrogate Chemically Convertible to Human Insulin." Angew. Chem. Int. Ed. Engl. 2010, 49 (32), 5489–5493.

(54) Sohma, Y., Kent, S. B. H. "Biomimetic Synthesis of Lispro Insulin via a Chemically Synthesized “Mini-Proinsulin” Prepared by Oxime-Forming Ligation." J. Am. Chem. Soc. 2009, 131 (44), 16313–16318.

(55) Royo, M., Alsina, J., Giralt, E., Slomcyznska, U., Albericio, F. "S- Phenylacetamidomethyl (Phacm): An Orthogonal Cysteine Protecting Group for Boc and Fmoc Solid-Phase Peptide Synthesis Strategies." J. Chem. Soc. [Perkin 1] 1995, No. 9, 1095.

(56) Góngora-Benítez, M., Basso, A., Bruckdorfer, T., Royo, M., Tulla-Puche, J., Albericio, F. "Eco-Friendly Combination of the Immobilized PGA Enzyme and the S -Phacm Protecting Group for the Synthesis of Cys- Containing Peptides." Chem. - Eur. J. 2012, 18 (50), 16166–16176.

(57) Shen, F., Zhang, Z.-P., Li, J.-B., Lin, Y., Liu, L. "Hydrazine-Sensitive Thiol Protecting Group for Peptide and Protein Chemistry." Org. Lett. 2011, 13 (4), 568–571.

(58) Adams, S. R., Tsien, R. Y. "Controlling Cell Chemistry with Caged Compounds." Annu. Rev. Physiol. 1993, 55 (1), 755–784.

(59) Chang, C., Niblack, B., Walker, B., Bayley, H. "A Photogenerated Pore- Forming Protein." Chem. Biol. 1995, 2, 391–400.

(60) Pan, P., Bayley, H. "Caged Cysteine and Thiophosphoryl Peptides." FEBS Lett. 1997, 405 (1), 81–85.

(61) Philipson, K. D., Gallivan, J. P., Brandt, G. S., Dougherty, D. A., Lester, H. A. "Incorporation of Caged Cysteine and Caged Tyrosine into a Transmembrane Segment of the Nicotinic ACh Receptor." Am. J. Physiol.-Cell Physiol. 2001, 281 (1), C195–C206.

(62) Smith, A. B., Savinov, S. N., Manjappara, U. V., Chaiken, I. M. "Peptide- Small Molecule Hybrids via Orthogonal Deprotection-Chemoselective Conjugation to Cysteine-Anchored Scaffolds. A Model Study." Org. Lett. 2002, 4 (23), 4041–4044.

(63) Erlich, L. A., Kumar, K. S. A., Haj-Yahya, M., Dawson, P. E., Brik, A. "N- Methylcysteine-Mediated Total Chemical Synthesis of Ubiquitin Thioester." Org. Biomol. Chem. 2010, 8 (10), 2392.

(64) Kotzur, N., Briand, B., Beyermann, M., Hagen, V. "Wavelength-Selective Photoactivatable Protecting Groups for Thiols." J. Am. Chem. Soc. 2009, 131 (46), 16927–16931.

(65) Klán, P., Šolomek, T., Bochet, C. G., Blanc, A., Givens, R., Rubina, M., Popik, V., Kostikov, A., Wirz, J. "Photoremovable Protecting Groups in Chemistry and Biology: Reaction Mechanisms and Efficacy." Chem. Rev. 2013, 113 (1), 119–191.

(66) Armishaw, C. J., Daly, N. L., Nevin, S. T., Adams, D. J., Craik, D. J., Alewood, P. F. "Selenoconotoxins, a New Class of Potent 7 Neuronal Nicotinic Receptor Antagonists." J. Biol. Chem. 2006, 281 (20), 14136– 14143.

(67) Guo, X., Shi, J., Tang, Z., Cui, D., Zhang, Y. "Synthesis and Biological Activity of Seleno Sunflower Trypsin Inhibitor Analog." Chem. Biol. Drug Des. 2006, 68 (6), 341–344.

(68) Holland-Nell, K., Meldal, M. "Maintaining Biological Activity by Using Triazoles as Disufide Bond Mimetics." Angew. Chem. Int. Ed. 2011, 50 (22), 5204–5206.

(69) Gori, A., Wang, C.-I. A., Harvey, P. J., Rosengren, K. J., Bhola, R. F., Gelmi, M. L., Longhi, R., Christie, M. J., Lewis, R. J., Alewood, P. F., Brust, A. "Stabilization of the Cysteine-Rich Conotoxin MrIA by Using a 1,2,3-Triazole as a Disulfide Bond Mimetic." Angew. Chem. Int. Ed. 2015, 54 (4), 1361–1364.

(70) Empting, M., Avrutina, O., Meusinger, R., Fabritz, S., Reinwarth, M., Biesalski, M., Voigt, S., Buntkowsky, G., Kolmar, H. “Triazole Bridge: Disulfide-Bond Replacement by Ruthenium-Catalyzed Formation of 1,5- Disubstituted 1,2,3-Triazoles." Angew. Chem. Int. Ed. 2011, 50 (22), 5207–5211.

(71) Williams, G. M., Lee, K., Li, X., Cooper, G. J. S., Brimble, M. A. "Replacement of the CysA7-CysB7 Disulfide Bond with a 1,2,3-Triazole Linker Causes Unfolding in Insulin Glargine." Org. Biomol. Chem. 2015, 13 (13), 4059–4063.

(72) Milenov, K., Barth, T., Jošt, K., Kasakov, L. "Effect of Deamino-Dicarba- Oxytocin and Oxytocin on Myoelectrical and Mechanical Activity of Uterus, Stomach and Small Intestine in Dog." Endocrinol. Exp. 1979, 13 (3), 177–183.

(73) DiMaio, J., Jaramillo, J., Wernic, D., Grenier, L., Welchner, E., Adams, J. "Synthesis and Biological Activity of Atrial Natriuretic Factor Analogues: Effect of Modifications to the Disulfide Bridge." J. Med. Chem. 1990, 33 (2), 661–667.

(74) Cuthbertson, A. S., Husbyn, M., Engebretsen, M., Hartmann, M., Lange, M., Sandosham, J., Fischer, P. M., Fjerdingstad, H., Løvhaug, D. "Design of Low Molecular Weight Hematoregulatory Agents from the Structure- Activity Relationship of a Dimeric Pentapeptide." J. Med. Chem. 1997, 40 (18), 2876–2882.

(75) Elaridi, J., Patel, J., Jackson, W. R., Robinson, A. J. "Controlled Synthesis of (S,S)-2,7-Diaminosuberic Acid: A Method for Regioselective Construction of Dicarba Analogues of Multicystine-Containing Peptides." J. Org. Chem. 2006, 71 (20), 7538–7545.

(76) MacRaild, C. A., Illesinghe, J., Lierop, B. J. van, Townsend, A. L., Chebib, M., Livett, B. G., Robinson, A. J., Norton, R. S. "Structure and Activity of

(2,8)-Dicarba-(3,12)-Cystino α-ImI, an α-Conotoxin Containing a Nonreducible Cystine Analogue." J. Med. Chem. 2009, 52 (3), 755–762.

(77) Cui, H.-K., Guo, Y., He, Y., Wang, F.-L., Chang, H.-N., Wang, Y.-J., Wu, F.-M., Tian, C.-L., Liu, L. "Diaminodiacid-Based Solid-Phase Synthesis of Peptide Disulfide Bond Mimics." Angew. Chem. Int. Ed. 2013, 52, 1–6.

(78) Hossain, M. A., Rosengren, K. J., Zhang, S., Bathgate, R. A. D., Tregear, G. W., van Lierop, B. J., Robinson, A. J., Wade, J. D. "Solid Phase Synthesis and Structural Analysis of Novel A-Chain Dicarba Analogs of Human Relaxin-3 (INSL7) That Exhibit Full Biological Activity." Org. Biomol. Chem. 2009, 7 (8), 1547.

(79) Zhang, S., Hughes, R. A., Bathgate, R. A. D., Shabanpoor, F., Hossain, M. A., Lin, F., van Lierop, B., Robinson, A. J., Wade, J. D. "Role of the Intra-A-Chain Disulfide Bond of Insulin-like Peptide 3 in Binding and Activation of Its Receptor, RXFP2." Peptides 2010, 31 (9), 1730–1736.

(80) Robinson, A., Van Lierop, B. Patent WO 2011146973 A1, "Insulin Analogues." WO 2011146973 A1.

(81) Spinella, M. J., Malik, A. B., Everitt, J., Andersen, T. T. "Design and Synthesis of a Specific Endothelin 1 Antagonist: Effects on Pulmonary Vasoconstriction." Proc. Natl. Acad. Sci. 1991, 88 (16), 7443–7446.

(82) Abraham, W. M., Ahmed, A., Cortes, A., Spinella, M. J., Malik, A. B., Andersen, T. T. "A Specific Endothelin-1 Antagonist Blocks Inhaled Endothelin-1-Induced Bronchoconstriction in Sheep." J. Appl. Physiol. 1993, 74 (5), 2537–2542.

(83) Thurieau, C., Janiak, P., Krantic, S., Guyard, C., Pillon, A., Kucharczyk, N., Vilaine, J. P., Fauchère, J. L. "A New Somatostatin Analog with Optimized Ring Size Inhibits Neointima Formation Induced by Balloon Injury in Rats without Altering Growth Hormone Release." Eur. J. Med. Chem. 1995, 30 (2), 115–122.

(84) Limal, D., Briand, J.-P., Dalbon, P., Jolivet, M. "Solid-Phase Synthesis and on-Resin Cyclization of a Disulfide Bond Peptide and Lactam Analogues Corresponding to the Major Antigenic Site of HIV gp41 Protein." J. Pept. Res. 1998, 52 (2), 121–129.

(85) Hargittai, B., Solé, N. A., Groebe, D. R., Abramson, S. N., Barany, G. "Chemical Syntheses and Biological Activities of Lactam Analogues of α- Conotoxin SI." J. Med. Chem. 2000, 43 (25), 4787–4792.

(86) Grieco, P., Gitu, P. M., Hruby, V. J. "Preparation of “side-Chain-to-Side- Chain” cyclic Peptides by Allyl and Alloc Strategy: Potential for Library Synthesis." J. Pept. Res. 2001, 57 (3), 250–256.

(87) Grieco, P., Carotenuto, A., Patacchini, R., Maggi, C. A., Novellino, E., Rovero, P. "Design, Synthesis, Conformational Analysis, and Biological Studies of Urotensin-II Lactam Analogues." Bioorg. Med. Chem. 2002, 10 (12), 3731–3739.

(88) Fázio, M. A., Oliveira, V. X., Bulet, P., Miranda, M. T. M., Daffre, S., Miranda, A. "Structure-Activity Relationship Studies of Gomesin: Importance of the Disulfide Bridges for Conformation, Bioactivities, and Serum Stability." Biopolymers 2006, 84 (2), 205–218.

(89) Büllesbach, E. E., Schwabe, C. "Replacement of Disulfides by Amide Bonds in the Relaxin-like Factor (RLF/INSL3) Reveals a Role for the A11– B10 Link in Transmembrane Signaling." Biochemistry (Mosc.) 2012, 51 (20), 4198–4205.

(90) Lebl, M., Hruby, V. J. "Synthesis of Cyclic Peptides by Solid Phase Methodology." Tetrahedron Lett. 1984, 25 (20), 2067–2068.

(91) Hidaka, Y., Ohmori, K., Wada, A., Ozaki, H., Ito, H., Hirayama, T., Takeda, Y., Shimonishi, Y. "Synthesis and Biological Properties of Carba- Analogs of Heat-Stable Enterotoxin (ST) Produced by Enterotoxigenic Escherichia Coli." Biochem. Biophys. Res. Commun. 1991, 176 (3), 958– 965.

(92) Mayer, J. P., Heil, J. R., Zhang, J., Munson, M. C. "An Alternative Solid- Phase Approach to C 1 -Oxytocin." Tetrahedron Lett. 1995, 36 (41), 7387–7390.

(93) Tugyi, R., Mezö, G., Fellinger, E., Andreu, D., Hudecz, F. "The Effect of Cyclization on the Enzymatic Degradation of Herpes Simplex Virus Glycoprotein D Derived Epitope Peptide." J. Pept. Sci. 2005, 11, 642–649.

(94) Galande, A. K., Bramlett, K. S., Burris, T. P., Wittliff, J. L., Spatola, A. F. "Thioether Side Chain Cyclization for Helical Peptide Formation: Inhibitors of Estrogen Receptor–coactivator Interactions." J. Pept. Res. 2004, 63 (3), 297–302.

(95) Knerr, P. J., Tzekou, A., Ricklin, D., Qu, H., Chen, H., van der Donk, W. A., Lambris, J. D. "Synthesis and Activity of Thioether-Containing Analogues of the Complement Inhibitor Compstatin." Chem. Biol. 2011, 6, 753–760.

(96) Bondebjerg, J., Grunnet, M., Jespersen, T., Meldal, M. "Solid-Phase Synthesis and Biological Activity of a Thioether Analogue of Conotoxin G1." ChemBioChem 2003, 4 (2-3), 186–194.

(97) Dekan, Z., Vetter, I., Daly, N. L., Craik, D. J., Lewis, R. J., Alewood, P. F. "α-Conotoxin ImI Incorporating Stable Cystathionine Bridges Maintains Full Potency and Identical Three-Dimensional Structure." J. Am. Chem. Soc. 2011, 133, 15866–15869.

(98) Jošt, K., Rudinger, J., Klostermeyer, H., Zahn, H. "Synthese Und Hypoglycämische Wirkung Eines Insulinanalogen Cystathionin-Peptides: Ein Argument Gegen Die Beteiligung Der Intrachenaren Disulfidgruppe Bei Der Insulin Wirkung." Z. Für Naturforschung B 1968, 23 b, 1059– 1061.

(99) Misra, S., Boylan, M., Selvam, A., Spallholz, J., Björnstedt, M. "Redox- Active Selenium Compounds—From Toxicity and Cell Death to Cancer Treatment." Nutrients 2015, 7 (5), 3536–3556.

Chapter 2

2-Nitroveratryl as a Photocleavable Thiol- Protecting Group for Directed Disulfide Bond Formation in the Chemical Synthesis of Insulin

Published as:

2.1 Abstract

Chemical synthesis of peptides can allow the option of sequential formation of multiple cysteines through exploitation of judiciously chosen regioselective thiol- protecting groups. Described herein is the use of 2-nitroveratryl (oNv) as a new orthogonal group that can be cleaved by photolysis under ambient conditions. In combination with complementary S-pyridinesulfenyl activation, disulfide bonds are formed rapidly in situ. The preparation of Fmoc-Cys(oNv)-OH is described together with its use for the solid-phase synthesis of complex cystine-rich peptides, such as insulin.

2.2 Introduction

Disulfide bonds are key structural elements in many therapeutically relevant peptides,1–3 and much work is being undertaken to optimize their pharmacological properties so that new lead compounds can be developed for preclinical evaluation. Typically, cystine-rich peptides are folded in a nondirected fashion in buffer at high pH by air oxidation because the most thermodynamically stable isomer often possesses the correct disulfide connectivity, particularly for native structures.4 However, when performing experiments such as alanine scanning for structure-activity relationship studies, a single mutation can significantly alter the folding pathway leading to misfolded isomers. In these instances, it is desirable to employ a regioselective chemical- synthesis strategy for peptides with multiple disulfide bonds to have certainty regarding the disulfide connectivity.

A number of orthogonal thiol-protecting groups which are SPPS compatible have been developed to efficiently form multiple cystine bonds in a stepwise manner.5 Synthetic protocols for disulfide-containing peptides, such as conotoxins, have been developed.6,7 The synthesis of multicystine heterodimeric peptides, such as insulin, represents a special challenge. The first reported multistep chemical assembly of this hormone employed S-Acm protection for the A7 – B7 cysteine pair, which was oxidatively cleaved by

36 treatment with elemental iodine to produce the correctly folded target peptide (Figure 1.5(ii)).8 Refinements of this methodology have been reported9–11 and applied to other insulin-like peptides.12,13 But modification of sensitive residues, such as methionine, tryptophan, and tyrosine, can occur when using the harsh reaction conditions therein described, leading to lower-than-expected yields.14 Therefore, the development of new S-protecting groups, which can be cleaved under ambient conditions, would be desirable to mitigate such outcomes. Enzyme-cleavable15 and hydrazine-cleavable16 protecting groups have been developed to satisfy this criterion. We propose the use of photocleavable protecting groups as a more attractive option for S-protection because they can be cleaved in aqueous media without the requirement of any reagents, thus minimizing the likelihood of side reactions. A range of photocleavable moieties has been in use since the 1960s, and have been primarily used as photocages for kinetic studies of various biomolecular interactions.17 Analogues of both the 2-nitrobenzyl (oNb) and oNv groups have been employed to mask thiols,18–20 and both N-α Boc21 and N-α Fmoc-protected22 Cys(oNb) building blocks are compatible with SPPS protocols. Further, oNb and coumaryl-based S-protecting groups have been incorporated into a bis-cysteinyl peptide to demonstrate wavelength-selective orthogonality.23 These examples highlight the versatility of photochemistry and the potential for its use as an orthogonal trigger. But to our knowledge, this technique has never been applied to the directed formation of disulfide bonds in synthetic peptides. We thus describe such an application.

Herein, oNv was preferred to oNb protection because it has a stronger molar absorptivity in the region of 350 nm. This is important because irradiation at shorter wavelengths can degrade photosensitive residues such as tryptophan and tyrosine.17 To form cystine bonds efficiently, an asymmetrical strategy was favored such that S-oNv and S-Pyr protection was used in combination,21 as depicted in Figure 2.1. The S-Pyr moiety is an excellent leaving group due to its relatively low pKa.24 During photolysis, the liberated thiolate can attack the “activated” sulfur atom resulting in formation of a disulfide bond plus two adducts. To demonstrate the efficacy and versatility of this approach, it was applied to the synthesis of a number of model peptides: oxytocin, α-conotoxin

ImI, and human insulin (Figures 2.2(i), (ii) and (iii)). Incorporation of the oNv- protected cysteine residue was initially achieved by an in situ on-resin method, but further investigation showed that a building block approach by using Fmoc- Cys(oNv)-OH to be more convenient.

2.3 Synthetic Methods

The synthesis of Fmoc-Cys(oNv)-OH (69% yield) was adapted from previously established protocols23 by using 2-nitroveratryl bromide and Fmoc-Cys-OH in the presence of diisopropylethylamine (DIEA) through an SN2 reaction mechanism. Cysteine residues are susceptible to racemization during activation prior to use in SPPS. Therefore, the coupling conditions were optimized to minimize this side reaction.25–27 The best combination of coupling reagents was found to be 1-hydroxybenzotriazole (HOBt) and diisopropylcarbodiimide (DIC)

38 activation, which led to <0.5% racemization after 5 min of preactivation (see Appendix 1, Section 9 for full details).

The neuropeptide oxytocin is a simple nonamer with a single disulfide bond and is hence an excellent model system (Figure 2.2(i)).28 The key objectives for this synthesis (Figure 2.3(i)) were to evaluate whether S-oNv protection is: (1) stable to 10% trifluoromethanesulfonic acid (TFMSA) in trifluoroacetic acid (TFA), which is a key step in the synthetic strategy for producing complex cystine-rich peptides, and (2) a viable method to form disulfide bonds efficiently by photolysis. tert-Butyl protection of the Cys1 residue (I) is necessary because this approach is intended for use where there are existing disulfide bonds present in a target peptide such as insulin. S-Pyr functionalization of this residue would result in disulfide shuffling with other thiols. The mono- methoxytrityl (Mmt) species at Cys6 was cleaved using 1% TFA in dichloromethane (DCM) followed by base-catalyzed reprotection with 2- nitroveratryl bromide to give the resin-bound species II, which was then isolated and purified by reversed-phase high-performance liquid chromatography (RP- HPLC). Cleavage of S-tBu and concomitant functionalization of the Cys1 thiol gave the S-Pyr intermediate (III), which was achieved by treatment with 10% TFMSA/TFA in the presence of excess dipyridine disulfide (DPDS), demonstrating that S-oNv is stable to these highly acidic conditions. Following purification, the peptide was dissolved in 20% aqueous acetonitrile at 0.5 mg/mL and irradiated at 350 nm for 30 minutes to give IV. The expected retention time and mass shifts were detected by analytical RP-HPLC and electrospray ionization mass spectrometry (ESI-MS; see Appendix 1, Section 11) indicating that the cyclization had occurred. This synthesis demonstrated that employing photocleavable S-protection is a robust and high-yielding method for forming disulfide bonds.

A similar photoprotection strategy was then applied to the synthesis of α- conotoxin ImI (Figures 2.2(ii) and 2.3(ii)), which is a selective antagonist of α7 nicotinic acetylcholine receptors and contains a characteristic 1-3/2-4 disulfide connectivity.29 In contrast to the in situ method for oxytocin, a building-block

39 approach was employed due to a lack of selectivity between the S-Mmt and S- Trt protecting groups during cleavage on the solid support (see Appendix 1, Section 12). After isolation of the linear precursor V, the first disulfide bond was formed by thiolysis to give VI and was followed by S-tBu cleavage and concomitant S-Pyr functionalization to give VII. α-Conotoxin ImI (VIII) was obtained by photolysis of VII at 350 nm for 20 min and isolated via RP-HPLC. The peptide was assayed for its binding activity to confirm that the correctly folded isomer was formed (experiments performed by Dr Irina Vetter and Prof. Richard Lewis).30 It was found that it concentration-dependently inhibited α7 nAChR-mediated Ca2+ responses with a half maximal inhibitory concentration

(IC50 ; 0.85 ± 0.13 mm; n = 5 experiments), which was comparable to that of the control peptide (IC50 1.07 ± 0.32 mm; n = 5 experiments). This further confirmed the efficacy of this methodology. The RP-HPLC, matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) and binding data are detailed in Appendix 1.

An analogous photoprotection strategy was applied to the synthesis of human insulin, which comprises of two interchain disulfide bonds connecting an A and a B chain as a parallel heterodimer, plus a third intrachain disulfide bond on the A chain (Figure 2.2(iii)). Due to its added complexity, two TFMSA/TFA/DPDS treatments were required to perform the synthesis. Initial attempts to assemble insulin whereby the A20-B19 cystine bond was to be formed last resulted in poor yields due to solubility problems of the A chain and significant disulfide shuffling during the final photolysis step (see Appendix 1, Section 14). The latter point suggests that the conformation of the penultimate intermediate was unfavorable for formation of the final disulfide bond indicating that the order in which they are formed is important. Therefore, an improved synthetic

41 methodology was devised (Figure 2.4) that featured two new design elements: (1) the A7 – B7 disulfide bond would be formed last, and (2) an iso-acyl dipeptide would be incorporated in to the A chain to enhance its solubility in aqueous media.31,32 After assembly and isolation of the A chain depsi-peptide (IX), the A6 – A11 intrachain disulfide bond was formed through thiolysis to give X followed by modification of the CysA20 residue to form XI in a similar fashion to that used for the α-conotoxin synthesis. Combination of the A and B chains produced the heterodimeric species XII and was followed by S-Pyr functionalization at CysB7 to form XIII with a yield of 50% post-purification.

After dissolution of XIII in 6 M guanidinium hydrochloride at 0.2 mg/mL, the photolysis reaction was complete after 45 minutes with one major product (XIV) detected by RP-HPLC. Conversion from the depsi-peptide to the native hormone XV was accomplished in quantitative yields by a pH adjustment from 5.5 to 7.5 (36% yield). The RP-HPLC chromatograms in Figure 2.5(i), (ii) and (iii) demonstrate that the key steps involved in the formation of the A7 – B7 disulfide bond were efficient. Figure 2.5(iv) confirms the identity of the isolated product by MALDI-MS. The total recovery to form the third disulfide bond was 18%, which is significantly higher than previously reported methods that use an iodine/di-Acm strategy (10.3% yield).12 Although non-directed folding methods have reported higher overall yields,33,34 this is an excellent result for a regioselective approach. The binding activity of the peptide was measured by Assoc. Prof. Briony Forbes in the insulin-receptor binding assay35 and was found to have the same affinity for the receptor as the positive control (Figure 2.5(v)), confirming that the native hormone had been successfully synthesized.

2.4 Conclusion

In summary, we have demonstrated the utility and versatility of 2-nitroveratryl as a thiol-protecting group, and have employed it as part of a broader strategy for the efficient synthesis of complex cystine-rich peptides. Fmoc-Cys(oNv)-OH was synthesized in good yield and incorporated without significant racemization.

The model peptides oxytocin and α-conotoxin ImI were successfully assembled, as was human insulin, which was synthesized in good yield. This methodology is fully compatible with Fmoc SPPS protocols including stability to TFMSA, which is useful when multiple levels of orthogonality are required. This versatile technique could be further enhanced by employing other photocleavable moieties such as coumaryls for wavelength-selective orthogonality and can also be applied to multistep native chemical-ligation syntheses for protein assembly.

2.5 Acknowledgments

This research was partially funded by the NHMRC Project grants 1023321 and 1023078 to M.A.H. and J.D.W. and C.S.I.R.O. (Australia). We thank the following people: Feng Lin for amino acid analysis, Dr. Keiko Hojo for assistance with the racemization study, Dr. Julien Tailhades for his expertise in iso-acyl dipeptides, Assoc. Prof. Uta Wille and Dr. Catrin Goeschen for access to their photoreactor, and Assoc. Prof. Craig Hutton and Dr. Jade Cottam for access to their polarimeter. J.D.W. is an NHMRC (Australia) Principal Research Fellow. Research at the Florey Institute of Neuroscience and Mental Health is supported by the Victorian Government Operational Infrastructure Support Program.

2.6 References

(1) Banting, F. G., Best, C. H., Collip, J. B., Campbell, W. R., Fletcher, A. A. "Pancreatic Extracts in the Treatment of Diabetes Mellitus." Can. Med. Assoc. J. 1922, 12 (3), 141.

(2) Miljanich, G. P. "Ziconotide: Neuronal Calcium Channel Blocker for Treating Severe Chronic Pain." Curr. Med. Chem. 2004, 11 (23), 3029– 3040.

(3) Teerlink, J. R., Metra, M., Felker, G. M., Ponikowski, P., Voors, A. A., Weatherley, B. D., Marmor, A., Katz, A., Grzybowski, J., Unemori, E., Teichman, S. L., Cotter, G. "Relaxin for the Treatment of Patients with Acute Heart Failure (Pre-RELAX-AHF): A Multicentre, Randomised, Placebo-Controlled, Parallel-Group, Dose-Finding Phase IIb Study." Lancet Lond. Engl. 2009, 373 (9673), 1429–1439.

(4) Andreu, D., Nicolas, E. "Disulfide Formation in Synthetic Peptides and Proteins: The State of the Art." In Solid-Phase Synthesis A Practical Guide, Marcel Dekker: New York, 2000, pp 365–376.

(5) Isidro-Llobet, A., Álvarez, M., Albericio, F. "Amino Acid-Protecting Groups." Chem. Rev. 2009, 109 (6), 2455–2504.

(6) Cuthbertson, A., Indrevoll, B. "A Method for the One-Pot Regioselective Formation of the Two Disulfide Bonds of α-Conotoxin SI." Tetrahedron Lett. 2000, 41 (19), 3661–3663.

(7) Cuthbertson, A., Indrevoll, B. "Regioselective Formation, Using Orthogonal Cysteine Protection, of an α-Conotoxin Dimer Peptide Containing Four Disulfide Bonds." Org. Lett. 2003, 5 (16), 2955–2957.

(8) Sieber, P., Kamber, B., Hartmann, A., Jöhl, A., Riniker, B., Rittel, W. "Totalsynthese von Humaninsulin. IV. Beschreibung Der Endstufen." Helv. Chim. Acta 1977, 60 (1), 27–37.

(9) 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 (24), 11384–11392.

(10) 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." Bioconjug. Chem. 2009, 20 (7), 1390–1396.

(11) Liu, F., Luo, E. Y., Flora, D. B., Mayer, J. P. "Concise Synthetic Routes to Human Insulin." Org. Lett. 2013, 15 (4), 960–963.

(12) Akhter Hossain, M., Bathgate, R. A. D., Kong, C. K., Shabanpoor, F., Zhang, S., Haugaard-Jönsson, 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 (11), 1816–1822.

(13) Chan, L. J., Smith, C. M., Chua, B. E., Lin, F., Bathgate, R. A. D., Separovic, F., Gundlach, A. L., Hossain, M. A., Wade, J. D. "Synthesis of Fluorescent Analogs of Relaxin Family Peptides and Their Preliminary in Vitro and in Vivo Characterization." Front. Chem. 2013, 1, 30.

(14) Kamber, B., Hartmann, A., Eisler, K., Riniker, B., Rink, H., Sieber, P., Rittel, W. "The Synthesis of Cystine Peptides by Iodine Oxidation of S- Trityl-Cysteine and S-Acetamidomethyl-Cysteine Peptides." Helv. Chim. Acta 1980, 63 (4), 899–915.

(15) Royo, M., Alsina, J., Giralt, E., Slomcyznska, U., Albericio, F. "S- Phenylacetamidomethyl (Phacm): An Orthogonal Cysteine Protecting Group for Boc and Fmoc Solid-Phase Peptide Synthesis Strategies." J. Chem. Soc. [Perkin 1] 1995, No. 9, 1095.

(16) Shen, F., Zhang, Z.-P., Li, J.-B., Lin, Y., Liu, L. "Hydrazine-Sensitive Thiol Protecting Group for Peptide and Protein Chemistry." Org. Lett. 2011, 13 (4), 568–571.

(17) Adams, S. R., Tsien, R. Y. "Controlling Cell Chemistry with Caged Compounds." Annu. Rev. Physiol. 1993, 55 (1), 755–784.

(18) Chang, C., Niblack, B., Walker, B., Bayley, H. "A Photogenerated Pore- Forming Protein." Chem. Biol. 1995, 2, 391–400.

(19) Pan, P., Bayley, H. "Caged Cysteine and Thiophosphoryl Peptides." FEBS Lett. 1997, 405 (1), 81–85.

(20) Philipson, K. D., Gallivan, J. P., Brandt, G. S., Dougherty, D. A., Lester, H. A. "Incorporation of Caged Cysteine and Caged Tyrosine into a Transmembrane Segment of the Nicotinic ACh Receptor." Am. J. Physiol.-Cell Physiol. 2001, 281 (1), C195–C206.

(21) Smith, A. B. 3rd, Savinov, S. N., Manjappara, U. V., Chaiken, I. M. "Peptide-Small Molecule Hybrids via Orthogonal Deprotection- Chemoselective Conjugation to Cysteine-Anchored Scaffolds. A Model Study." Org. Lett. 2002, 4 (23), 4041–4044.

(22) Erlich, L. A., Kumar, K. S. A., Haj-Yahya, M., Dawson, P. E., Brik, A. "N- Methylcysteine-Mediated Total Chemical Synthesis of Ubiquitin Thioester." Org. Biomol. Chem. 2010, 8 (10), 2392.

(23) Kotzur, N., Briand, B., Beyermann, M., Hagen, V. "Wavelength-Selective Photoactivatable Protecting Groups for Thiols." J. Am. Chem. Soc. 2009, 131 (46), 16927–16931.

(24) Akaji, K., Kiso, Y. "Synthesis of Cystine Peptides." In Houben-Weyl Synthesis of Peptides and Peptidomimetics, Thieme Publishing Group: Stuttgart, 2002, Vol. E 22b, pp 101–141.

(25) Han, Y., Albericio, F., Barany, G. "Occurrence and Minimization of Cysteine Racemization during Stepwise Solid-Phase Peptide Synthesis." J. Org. Chem. 1997, 62 (13), 4307–4312.

(26) Angell, Y. M., Alsina, J., Barany, G., Albericio, F. "Practical Protocols for Stepwise Solid-Phase Synthesis of Cysteine-Containing Peptides." J. Pept. Res. 2002, 60 (5), 292–299.

(27) Hojo, K., Shinozaki, N., Hara, A., Onishi, M., Fukumori, Y., Ichikawa, H. "Aqueous Microwave-Assisted Solid-Phase Peptide Synthesis Using Fmoc Strategy. II. Racemization Studies and Water Based Synthesis of Cysteine-Containing Peptides." Protein Pept. Lett. 2013, 20 (10), 1122– 1128.

(28) Carson, D. S., Guastella, A. J., Taylor, E. R., McGregor, I. S. "A Brief History of Oxytocin and Its Role in Modulating Psychostimulant Effects." J. Psychopharmacol. Oxf. Engl. 2013, 27 (3), 231–247.

(29) McIntosh, J. M., Yoshikami, D., Mahe, E., Nielsen, D. B., Rivier, J. E., Gray, W. R., Olivera, B. M. "A Nicotinic Acetylcholine Receptor Ligand of Unique Specificity, Alpha-Conotoxin ImI." J. Biol. Chem. 1994, 269 (24), 16733–16739.

(30) Vetter, I., Lewis, R. J. "Characterization of Endogenous Calcium Responses in Neuronal Cell Lines." Biochem. Pharmacol. 2010, 79 (6), 908–920.

(31) Yoshiya, T., Taniguchi, A., Sohma, Y., Fukao, F., Nakamura, S., Abe, N., Ito, N., Skwarczynski, M., Kimura, T., Hayashi, Y., Kiso, Y. "O-Acyl Isopeptide Method For Peptide Synthesis: Synthesis of Forty Kinds of O - Acyl Isodipeptide Unit Boc-Ser/Thr(Fmoc-Xaa)-OH." Org. Biomol. Chem. 2007, 5 (11), 1720.

(32) Liu, F., Luo, E. Y., Flora, D. B., Mezo, A. R. "A Synthetic Route to Human Insulin Using Isoacyl Peptides." Angew. Chem. Int. Ed Engl. 2014, 53 (15), 3983–3987.

(33) Sohma, Y., Kent, S. B. H. "Biomimetic Synthesis of Lispro Insulin via a Chemically Synthesized “Mini-Proinsulin” Prepared by Oxime-Forming Ligation." J. Am. Chem. Soc. 2009, 131 (44), 16313–16318.

(34) Sohma, Y., Hua, Q.-X., Whittaker, J., Weiss, M. A., Kent, S. B. H. "Design and Folding of [GluA4(OβThrB30)]Insulin (“Ester Insulin”), a Minimal Proinsulin Surrogate Chemically Convertible to Human Insulin." Angew. Chem. Int. Ed. Engl. 2010, 49 (32), 5489–5493.

(35) Denley, A., Bonython, E. R., Booker, G. W., Cosgrove, L. J., Forbes, B. E., Ward, C. W., Wallace, J. C. "Structural Determinants for High-Affinity Binding of Insulin-Like Growth Factor II to Insulin Receptor (IR)-A, the Exon 11 Minus Isoform of the IR." Mol. Endocrinol. 2004, 18 (10), 2502– 2512.

Chapter 3

Total Chemical Synthesis of a Heterodimeric Interchain Bis-Lactam-Linked Peptide: Application to an Analogue of Human Insulin-Like Peptide 3

Published as:

3.1 Abstract

Nonreducible cystine isosteres represent important peptide design elements in that they can maintain a near-native tertiary conformation of the peptide while simultaneously extending the in vitro and in vivo half-life of the biomolecule. Examples of these cystine mimics include dicarba, diselenide, thioether, triazole, and lactam bridges. Each has unique physicochemical properties that impact upon the resulting peptide conformation. Each also requires specific conditions for its formation via chemical peptide synthesis protocols. While the preparation of peptides containing two lactam bonds within a peptide is technically possible and reported by others, to date there has been no report of the chemical synthesis of a heterodimeric peptide linked by two lactam bonds. To examine the feasibility of such an assembly, judicious use of a complementary combination of amine and acid protecting groups together with nonfragment-based, total stepwise solid phase peptide synthesis led to the successful preparation of an analogue of the model peptide, INSL3, in which both of the interchain disulfide bonds were replaced with a lactam bond. An analogue containing a single disulfide-substituted interchain lactam bond was also prepared. Both INSL3 analogues retained significant cognate relaxin/insulin-like family peptide receptor 2 (RXFP2) binding affinity.

3.2 Introduction

Cysteine-rich peptides such as conotoxins and insulin-like peptides are an increasingly important class of biomolecules. They usually possess intricately folded, sometimes knotted, structures and some have been developed as treatments for a variety of conditions, such as pain,1,2 cancer,3 diabetes mellitus4 and heart failure.5,6 As such, much work is being undertaken to optimize their pharmacological properties so that new lead compounds are developed for preclinical evaluation. Disulfide bonds play a critical role in maintaining the peptide conformation and biological activity of these molecules. However, they are susceptible to reduction in vivo, as part of the normal degradative process which, in turn, can disrupt the three-dimensional structure

52 and lead to loss of activity. In order to stabilize peptide structures, numerous disulfide bond mimics have been developed. Guo et al. substituted a diselenide for a disulfide bond in a sunflower trypsin inhibitor which retained high potency.7 Armishaw et al. also applied this further to an α-conotoxin, which maintained full biological activity and had enhanced stability under biologically reducing conditions.8 This same model peptide was also prepared with thioether bonds as cystine mimics, and a similar outcome was achieved with respect to both activity and stability.9 Further, Holland-Nell and Meldal reported that 1,4- triazoles using the copper(I)-catalyzed azide-alkyne cycloaddition can also be a useful cystine isostere.10 Further work in this area has demonstrated that 1,5- triazoles, produced via ruthenium(II) catalysis, can be an even more effective mimic.11 Numerous dicarba analogues of cystine-containing peptides have also been prepared12–15 and shown to possess near-native structure and extended in vivo stability. Mono-substituted dicarba bond analogues of the heterodimeric peptides relaxin-316 and INSL317 have also been prepared and evaluated.

The incorporation of lactam bridges in peptides has been widely reported in the literature. Such linkages have been employed as both “staples” in order to stabilize α-helices and other secondary structures18–20 and as a strategy for generating small cyclic libraries.21 They have also found use as stable cystine isosteres21–28 as an alternative to the previous methods described. These linkages have two key structural characteristics: (1) there are dual orientations of the asymmetrical amide bond, and (2) bridge length can vary. The direction of the lactam bond can have either a negligible effect on binding and biological activity22 or a dramatic one. Interestingly, Hargittai and coworkers found that an α-conotoxin analogue with a Lys/Glu lactam bond was 4,000 times more potent than the “inverted” Glu/Lys analogue.23 Different pairings of side-chain carboxylate (Asp, Glu) and amine (Dpr, Dab, Orn, Lys) residues will of course vary the length of the lactam bridge which can also affect peptide conformation and activity.24–26 The most common residues used are an aspartyl/2,3- diaminopropionyl pairing, since the resultant side-chain to side-chain amide bond will result in the same number of atoms as a cystine bond.

There have been a number of methods developed for synthesizing lactam- containing peptides.22,26–28 Earlier work focused on a Boc/Bzl-based strategy, employing the base-labile O-Fm and Fmoc protecting groups for orthogonal protection of carboxylate and amine side-chains, respectively. Typically, the lactam is formed on the solid support, followed by HF cleavage and RP-HPLC. This methodology has been extended to assemble bis-lactam analogues18 including peptides with overlapping lactam bonds.19 The latter was performed by employing hydrazine-labile O-Dmab and Dde protecting groups in addition to the 9-fluorenylmethyl-based ones as semi-orthogonal protection. Thurieau et al. devised a solution-phase approach whereby the N-terminus and a lysine residue were protected until after lactamization in order to yield one discrete product.24 Allyl/Alloc protection is a common strategy when using an “on-resin” Fmoc/tBu approach21,23,24,28 as is the extremely acid labile and more convenient O-2-PhiPr/Mtt pair.20,29

INSL3 was chosen as the model system to evaluate lactam bonds as cystine isosteres in complex peptide structures. INSL3 is a hormone which has been shown to play an important role in testicular descent during sexual development.30 It is heterodimeric in nature with a 26-residue A-chain and a 31- residue B-chain. It contains a disulfide bonding configuration, that is, characteristic of the insulin/relaxin Superfamily31 with one intra-A-chain and two interchain (between A and B) cystine bonds which stabilize the three- dimensional structure of the peptide (Figure 3.1(i)). Its structure-activity relationship has been studied in detail, and both the A- and B-chains are required for RXFP2 receptor activation.32–34 Further, an analogue of the INSL3 B-chain alone was shown to be a potent RXFP2 antagonist.35 Recently, Büllesbach and Schwabe prepared analogues of human INSL3 in which one or the other of the two interchain disulfide bonds were replaced with a lactam bond, in which the purpose was to study the role of the native cystine (A11 – B10) and (A24 – B22) bonds in both binding and receptor activation.29 Their synthetic strategy involved native chemical ligation of two A-chain fragments followed by a second ligation with the B-chain. While very effective, the methodology requires the preparation of peptide thioesters as ligation

54 intermediates prior to subsequent off-resin preparation of the mono-intrachain lactam INSL3 analogue.

Moreover, to date, there have been no reports of a chemical synthesis of a bis- lactam cross-linked heterodimeric peptide. Consequently, in this study, we undertook an examination of the feasibility of a total step-wise synthesis of, first, a mono-lactam A24-B22 analogue and, subsequently, a bis-lactam A11 – B10/A24 – B22 analogue of INSL3. Our novel strategy involved the continuous assembly of both the A and B-chains, plus the lactam bridges on the same solid support, that is, without using preformed peptide fragments. In order to simplify assembly, the C-terminus of the A-chain was truncated at position 24, omitting ProA25 and TyrA26 that results in the analogue ΔA25/26 human INSL3 (Figure 3.1(ii)). Although receptor activation is somewhat diminished for this particular analogue, it still demonstrates significant binding affinity compared to that of the native form32 and hence was deemed to be a suitable model system to evaluate the synthetic feasibility of assembling these lactam analogues. Described herein is the synthesis of ΔA25/26 human INSL3 having a lactam substitution of CysA24,B10 as well as a first ever reported assembly of a heterodimeric analogue with two interchain lactam bridges, namely, ΔA25/26 human INSL3 bis-lactam A11 – B10/A24 – B22. The cognate RXFP2 G protein-coupled receptor binding affinity of the analogues was also evaluated.

3.3 Materials and Methods

Solid Phase Peptide Synthesis. All peptides were assembled on a CEM Liberty microwave peptide synthesizer (DKSH, Australia), except for when an unusual amino acid derivative was used whereby it was coupled manually. Standard Fmoc amino acids were obtained from GL Biochem (China) as was HATU and Boc anhydride. Fmoc-L-Asp-OtBu, Fmoc-L-Dpr(ivDde)-OH, Fmoc-L-Dpr(Mmt)- OH, and Fmoc-L-Asp(O-2-PhiPr)-OH were sourced from Novabiochem (Australia). Piperidine, hydrazine, DIEA, NMP, TIPS, DODT, TFA, iodine, and DOWEX ion exchange resin were obtained from Sigma-Aldrich (Australia). DCM and DMF were purchased from Ajax Finechem P/L (Australia). Fmoc-L- Ala-PEG-PS resin with a substitution of 0.2mmol/g was obtained from Applied Biosystems (Australia).

RP-HPLC Purification and Analysis. All peptides were analyzed on an Agilent 1100 Series HPLC (Australia) with an Agilent Eclipse XDB-C18 column (5 µm, 4.6 × 150 mm). The buffer system used was 0.1% TFA in water (buffer A) and 0.1% TFA in acetonitrile (buffer B). A typical gradient was 0–80% buffer B over 40 minutes at a 1 mL/min flow rate with detection at 220 nm. Purification was carried out on an Agilent 1200 Series HPLC using either an Agilent Eclipse XDB-C18 column (5 µm, 9.4 mm × 250 mm) or a Phenomenex Synergi Hydro C18 column (4 µm, 21.2 mm × 50mm). A typical gradient of 0–60% buffer B over 60 minutes at a flow rate of 5 mL/min was used.

Mass Spectrometry. The linear and intermediate peptides were characterized on an Agilent 6510 Dual ESI QTOF mass spectrometer (Australia).

Receptor Binding Assay. The receptor binding affinity of both analogues was determined in HEK-293T cells stably transfected with RXFP2 as previously described.32 Briefly, a single concentration of europium-labelled INSL3 (0.3 nM)36 was used in the presence of increasing concentration of the unlabelled INSL3 and the analogues (0.01 nM–1 µM). The data was analyzed using GraphPad PRISM 4 (GraphPad Inc., San Diego, USA) and expressed as mean

± SEM of three independent experiments. The statistical differences in pKi

56 values were calculated using one-way ANOVA coupled to Bonferroni’s multiple comparison test for multiple group comparisons.

Synthesis of ΔA25/26 Human INSL3 Mono-lactam A24-B22. INSL3 B-chain was assembled on Fmoc-L-Ala-PEG PS resin via microwave-assisted SPPS on a 50 µmol scale using an Fmoc SPPS approach (Figure 3.2). ivDde-protected diaminopropionic acid (Dpr) was used in place of CysB22, and CysB10 was incorporated with Acm protection. After chain elongation, the N-terminus of the resin-bound peptide was Boc-protected using di-tert-butyl carbonate in the presence of DIEA, followed by treatment with a solution of 3% hydrazine (5 × 3 minutes) to cleave the ivDde group. The free amine at the side-chain was then acylated via the HATU activated side-chain of Fmoc-L-Asp-OtBu, and the A- chain was then assembled on the same resin. CysA11 was also incorporated with Acm protection. The peptide was then TFA-cleaved, isolated, and then RP- HPLC-purified. The free sulfhydryls at residues CysA10 and CysA15 were then oxidized with 2 equivalents of iodine in a 50% acetic acid aqueous solution at a concentration of 1 mg/mL for 30 minutes. Excess iodine was quenched with DOWEX ion exchange resin, followed by a second RP-HPLC purification. The CysA11 – CysB10 intermolecular disulfide bond was then formed via oxidative cleavage of both Cys(Acm) residues using 20 equivalents of iodine.37 After a similar workup to the previous oxidation, the crude product was RP-HPLC purified to a high level and characterized via ESI-MS which gave a molecular mass of 6011 Da (theory: 6011.0). The final mass recovery was 700 µg (yield 1.8% relative to starting crude cleaved peptide).

Synthesis of ΔA25/26 Human INSL3 Bis-Lactam A11-B10/A24-B22. A similar synthetic strategy was employed for this synthesis on the same 50 µmol scale; however, the lactam bridge mimicking the CysA11 – CysB10 bond was also formed on the solid support. Thus B-chain was assembled using Dpr(Mmt) and Dpr(ivDde) at positions B10 and B22, respectively. His(Boc) was used instead of His(Trt) at positions B12 and B13 and was incorporated manually under ambient conditions using HOBt/DIC activation. After N-terminal capping and ivDde cleavage, the A-chain was assembled up to position A11 using an Asp(O-

2-PhiPr) residue and leaving the N-terminus Fmoc protected. At positions A17 and A18, unprotected Gln was incorporated (via HOBt/DIC activation) instead of the usual Trt side-chain protected derivative. The resin was then treated with 1% TFA (10 × 3 minutes) to cleave the O-2-PhiPr and Mmt protecting groups from the Asp and Dpr side-chains, respectively. After neutralization with mild base, the resin was treated with 1.5 equivalents of HATU in the presence of DIEA to form the A11 – B10 interchain amide bond. A small-scale pilot cleavage was performed in order to determine the success of this on-resin cyclization. After reaction, the remainder of the A-chain was assembled and the peptide was isolated and purified via RP-HPLC. Oxidation of the CysA10 – CysA15 pair was performed using 2 equivalents of elemental iodine as described above. The product was repurified and lyophilized. 20 µg of material at high purity was recovered (0.08% relative to the isolated crude material), and the peptide was characterized via ESI-MS which showed a molecular mass of 5990.0 Da (theory: 5989.8).

3.4 Results and Discussion

An initial assembly of the ΔA25/26 human INSL3 mono-lactam A24 – B22 was undertaken via total chemical synthesis (Figure 3.2). This analogue was chosen over the ΔA25/26 human INSL3 mono-lactam A11 – B10 analogue solely for reasons of synthetic simplicity with the former having its lactam bond closest to the C-terminus of the synthesis and being theoretically easier to form first before steric crowding became too great a consideration. The microwave- assisted assembly was successful as indicated by RP-HPLC and ESI-MS analysis of the crude S-reduced peptide (Figure 3.3(i)). The principal impurities were identified to be post-cleavage tert-butyl adducts which are characteristic of thiol and thioether-containing peptides. After RP-HPLC purification, mild oxidative conditions were employed to form the first A-chain intramolecular (A10 – A15) disulfide bond, followed by a second purification. The final oxidation step to form the A11 – B10 disulfide bond gave a number of side products which is a common outcome from treating peptides with a large

59 excess of iodine.37 Trp, Tyr, and Met residues can be modified, all of which are present in INSL3. Furthermore, misfolded isomers and dimers can occur due to disulfide shuffling. Because of this complex mixture, a two-step purification was performed, firstly using a conventional C18 Agilent column, followed by a second C18 Phenomenex column with polar residues bonded to the stationary phase. The rationale behind this strategy was to achieve an alternative selectivity from the chromatography such that most impurities coeluting with the parent compound during the first isolation will be separable during the second pass through the column. This strategy yielded sufficient peptide in high purity (Figures 3.3(ii) and 3.3(iii)) which enabled the analogue to be evaluated in the RXFP2 receptor-binding assay.

Assembly of the ΔA25/26 human INSL3 bis-lactam A11 – B10/A24 – B22 analogue B-chain (Figure 3.4) also proceeded smoothly as verified by a small- scale cleavage of the peptide resin. After ivDde deprotection at DprB22, the A- chain was assembled up to Fmoc-AspA11. RP-HPLC analysis indicated that the target intermediate was the major product. It was thought that forming the intermolecular lactam bond at this point was preferable, since extending the A- chain beforehand would increase the steric crowding on the resin, making it difficult to drive the lactamization to completion. Using the same rationale, Boc- protected HisB11 and HisB12 and unprotected GlnA17 and GlnA18 were also used, instead of the bulkier trityl-based derivatives. Deprotection of AspA11 and DprB10 with dilute TFA was monitored colorimetrically via Trt cleavage; however, it was not possible to determine whether full removal of the O-2-PhiPr moiety occurred. A trial TFA deprotection showed that this protecting group would cleave off regardless. After neutralization of the resin, the amide bond between residues A11 and B10 was formed. RP-HPLC analysis found that there were a large number of side products generated (Figure 3.5(i)), some of which were likely to be oligomeric species caused by undesired cross-linking within the solid support. Nevertheless, the last ten residues were coupled under microwave- assisted conditions, followed by cleavage and deprotection. As expected, the crude material consisted of multiple products and a two-step purification was again employed as described above in order to isolate the semipure reduced

61 species. Treatment with two equivalents of iodine gave the desired product, and a second two-step purification was performed, due to the complexity of the mixture. Despite the low yield, sufficient purified peptide was isolated (Figures 3.5(ii) and 3.5(iii)) for evaluation in the binding assay.

Both the ΔA25/26 human INSL3 mono-lactam A24-B22 and bis-lactam A11 – B10/A24 – B22 analogues were tested in the RXFP2 competition binding assay by Dr Fazel Shabanpoor (Figure 3.6). The receptor binding affinity (pKi) of each was found to be 8.35 ± 0.11 and 7.92 ± 0.12, respectively. While these values were approximately 10-fold and 12-fold lower compared to native INSL3 (9.24 ± 0.02), the more relevant comparison is with ΔA25,26 human INSL3 (8.59 ± 0.06), which shows that both analogues are only 1.7-fold and 4.6-fold lower. This showed that employing lactam bridges as stable cystine isosteres in insulin-like peptides holds much promise. The very modest reduction in binding affinity of ΔA25/26 human INSL3 mono-lactam A24 – B22 compared to ΔA25/26 human INSL3 is consistent with similar observations recently made for synthetic human INSL3 mono-lactam A24 – B22 versus native INSL3.29 The reason for this is probably due to a subtle change in the conformation that is caused by one or more of the following: the shorter length of the amide bond compared to the disulfide bond, change in the electronic nature of the cystine mimic, and steric effects. Unfortunately, insufficient material was obtained to perform structural studies using CD or NMR spectroscopy.38,39 It was also not possible to perform cAMP activity-based assays or, equally important, serum

62 stability assays in order to verify whether the more stable amide bond(s) gives the model peptides enhanced stability. Yet, importantly, this study demonstrates the feasibility of acquiring such heterodimeric analogues, although further work is clearly required to improve their total synthesis and subsequent yields of these analogues. This will include employing different protecting group combinations (e.g., Allyl/Alloc), incorporating a solution-phase step,40 or using a very low loading resin to minimize both steric crowding during assembly and the non-specificity of the on-resin directed lactam bond formation.41

3.5 Conclusion

In conclusion, a novel “two-chain assembly” strategy on a solid support was developed for synthesizing complex bis-lactam interchain-linked heterodimeric insulin-like peptides for the first time. This was achieved by use of microwave- assisted SPPS and use of semiorthogonal side-chain protecting groups. This approach enabled the preparation of both mono- and bis-lactam analogues of ΔA25/26 human INSL3 which were subsequently shown to have significant binding affinity at nanomolar concentration for RXFP2 receptor.

3.6 Acknowledgments

The authors thank the Bio21 Institute, University of Melbourne, for supporting this work. This research was also partially funded by NHMRC (Australia) project Grant 508995 and 1023078 to JDW. They thank Feng Lin for amino acid analysis and A/Prof RAD Bathgate (Florey Institute of Neuroscience and Mental Health) for access to the in vitro RXFP2 bioassay. Fazel Shabanpoor is an NHMRC CJ Martin Fellow and John D. Wade is an NHMRC (Australia) Principal Research Fellow. Research at the Florey Institute of Neuroscience and Mental Health is supported by the Victorian Government Operational Infrastructure Support Program.

3.7 References

(1) Satkunanathan, N., Livett, B., Gayler, K., Sandall, D., Down, J., Khalil, Z. "Alpha-Conotoxin Vc1.1 Alleviates Neuropathic Pain and Accelerates Functional Recovery of Injured Neurones." Brain Res. 2005, 1059 (2), 149–158.

(2) Miljanich, G. P. "Ziconotide: Neuronal Calcium Channel Blocker for Treating Severe Chronic Pain." Curr. Med. Chem. 2004, 11 (23), 3029– 3040.

(3) Ma, M. T., Karas, J. A., White, J. M., Scanlon, D., Donnelly, P. S. "A New Bifunctional Chelator for Copper Radiopharmaceuticals: A Cage Amine Ligand with a Carboxylate Functional Group for Conjugation to Peptides." Chem. Commun. Camb. Engl. 2009, No. 22, 3237–3239.

(4) Banting, F. G., Best, C. H., Collip, J. B., Campbell, W. R., Fletcher, A. A. "Pancreatic Extracts in the Treatment of Diabetes Mellitus." Can. Med. Assoc. J. 1922, 12 (3), 141.

(5) Teerlink, J. R., Metra, M., Felker, G. M., Ponikowski, P., Voors, A. A., Weatherley, B. D., Marmor, A., Katz, A., Grzybowski, J., Unemori, E., Teichman, S. L., Cotter, G. "Relaxin for the Treatment of Patients with Acute Heart Failure (Pre-RELAX-AHF): A Multicentre, Randomised, Placebo-Controlled, Parallel-Group, Dose-Finding Phase IIb Study." Lancet Lond. Engl. 2009, 373 (9673), 1429–1439.

(6) Teerlink, J. R., Cotter, G., Davison, B. A., Felker, G. M., Filippatos, G., Greenberg, B. H., Ponikowski, P., Unemori, E., Voors, A. A., Adams, K. F. J., Dorobantu, M. I., Grinfeld, L. R., Jondeau, G., Marmor, A., Masip, J., Pang, P. S., Werdan, K., Teichman, S. L., Trapani, A., Bush, C. A., Saini, R., Schumacher, C., Severin, T. M., Metra, M. "Serelaxin, Recombinant Human Relaxin-2, for Treatment of Acute Heart Failure (RELAX-AHF): A Randomised, Placebo-Controlled Trial." Lancet Lond. Engl. 2013, 381 (9860), 29–39.

(7) Guo, X., Shi, J., Tang, Z., Cui, D., Zhang, Y. "Synthesis and Biological Activity of Seleno Sunflower Trypsin Inhibitor Analog." Chem. Biol. Drug Des. 2006, 68 (6), 341–344.

(8) Armishaw, C. J., Daly, N. L., Nevin, S. T., Adams, D. J., Craik, D. J., Alewood, P. F. "Selenoconotoxins, a New Class of Potent 7 Neuronal Nicotinic Receptor Antagonists." J. Biol. Chem. 2006, 281 (20), 14136– 14143.

(9) Dekan, Z., Vetter, I., Daly, N. L., Craik, D. J., Lewis, R. J., Alewood, P. F. "α-Conotoxin ImI Incorporating Stable Cystathionine Bridges Maintains Full Potency and Identical Three-Dimensional Structure." J. Am. Chem. Soc. 2011, 133, 15866–15869.

(10) Holland-Nell, K., Meldal, M. "Maintaining Biological Activity by Using Triazoles as Disufide Bond Mimetics." Angew. Chem. Int. Ed. 2011, 50 (22), 5204–5206.

(11) Empting, M., Avrutina, O., Meusinger, R., Fabritz, S., Reinwarth, M., Biesalski, M., Voigt, S., Buntkowsky, G., Kolmar, H. “Triazole Bridge: Disulfide-Bond Replacement by Ruthenium-Catalyzed Formation of 1,5- Disubstituted 1,2,3-Triazoles." Angew. Chem. Int. Ed. 2011, 50 (22), 5207–5211.

(12) Elaridi, J., Patel, J., Jackson, W. R., Robinson, A. J. "Controlled Synthesis of (S,S)-2,7-Diaminosuberic Acid: A Method for Regioselective Construction of Dicarba Analogues of Multicystine-Containing Peptides." J. Org. Chem. 2006, 71 (20), 7538–7545.

(13) Cuthbertson, A. S., Husbyn, M., Engebretsen, M., Hartmann, M., Lange, M., Sandosham, J., Fischer, P. M., Fjerdingstad, H., Løvhaug, D. "Design of Low Molecular Weight Hematoregulatory Agents from the Structure- Activity Relationship of a Dimeric Pentapeptide." J. Med. Chem. 1997, 40 (18), 2876–2882.

(14) MacRaild, C. A., Illesinghe, J., Lierop, B. J. van, Townsend, A. L., Chebib, M., Livett, B. G., Robinson, A. J., Norton, R. S. "Structure and Activity of (2,8)-Dicarba-(3,12)-Cystino α-ImI, an α-Conotoxin Containing a Nonreducible Cystine Analogue." J. Med. Chem. 2009, 52 (3), 755–762.

(15) DiMaio, J., Jaramillo, J., Wernic, D., Grenier, L., Welchner, E., Adams, J. "Synthesis and Biological Activity of Atrial Natriuretic Factor Analogues: Effect of Modifications to the Disulfide Bridge." J. Med. Chem. 1990, 33 (2), 661–667.

(16) Hossain, M. A., Rosengren, K. J., Zhang, S., Bathgate, R. A. D., Tregear, G. W., van Lierop, B. J., Robinson, A. J., Wade, J. D. "Solid Phase Synthesis and Structural Analysis of Novel A-Chain Dicarba Analogs of Human Relaxin-3 (INSL7) That Exhibit Full Biological Activity." Org. Biomol. Chem. 2009, 7 (8), 1547.

(17) Zhang, S., Hughes, R. A., Bathgate, R. A. D., Shabanpoor, F., Hossain, M. A., Lin, F., van Lierop, B., Robinson, A. J., Wade, J. D. "Role of the Intra-A-Chain Disulfide Bond of Insulin-like Peptide 3 in Binding and Activation of Its Receptor, RXFP2." Peptides 2010, 31 (9), 1730–1736.

(18) Koerber, S. C., Gulyas, J., Lahrichi, S. L., Corrigan, A., Craig, A. G., Rivier, C., Vale, W., Rivier, J. "Constrained Corticotropin-Releasing Factor

(CRF) Agonists and Antagonists with i−(i+3) Glu-Xaa-DXbb-Lys Bridges." J. Med. Chem. 1998, 41 (25), 5002–5011.

(19) Lanigan, M. D., Pennington, M. W., Lefievre, Y., Rauer, H., Norton, R. S. "Designed Peptide Analogues of the Potassium Channel Blocker ShK Toxin." Biochemistry (Mosc.) 2001, 40 (51), 15528–15537.

(20) Dong, M., Te, J. A., Xu, X., Wang, J., Pinon, D. I., Storjohann, L., Bordner, A. J., Miller, L. J. "Lactam Constraints Provide Insights into the Receptor- Bound Conformation of Secretin and Stabilize a Receptor Antagonist." Biochemistry (Mosc.) 2011, 50 (38), 8181–8192.

(21) Grieco, P., Gitu, P. M., Hruby, V. J. "Preparation of “Side-Chain-to-Side- Chain” Cyclic Peptides by Allyl and Alloc Strategy: Potential for Library Synthesis." J. Pept. Res. 2001, 57 (3), 250–256.

(22) Limal, D., Briand, J.-P., Dalbon, P., Jolivet, M. "Solid-Phase Synthesis and on-Resin Cyclization of a Disulfide Bond Peptide and Lactam Analogues Corresponding to the Major Antigenic Site of HIV gp41 Protein." J. Pept. Res. 1998, 52 (2), 121–129.

(23) Hargittai, B., Solé, N. A., Groebe, D. R., Abramson, S. N., Barany, G. "Chemical Syntheses and Biological Activities of Lactam Analogues of α- Conotoxin SI." J. Med. Chem. 2000, 43 (25), 4787–4792.

(24) Thurieau, C., Janiak, P., Krantic, S., Guyard, C., Pillon, A., Kucharczyk, N., Vilaine, J. P., Fauchère, J. L. "A New Somatostatin Analog with Optimized Ring Size Inhibits Neointima Formation Induced by Balloon Injury in Rats without Altering Growth Hormone Release." Eur. J. Med. Chem. 1995, 30 (2), 115–122.

(25) Grieco, P., Carotenuto, A., Patacchini, R., Maggi, C. A., Novellino, E., Rovero, P. "Design, Synthesis, Conformational Analysis, and Biological Studies of Urotensin-II Lactam Analogues." Bioorg. Med. Chem. 2002, 10 (12), 3731–3739.

(26) Fázio, M. A., Oliveira, V. X., Bulet, P., Miranda, M. T. M., Daffre, S., Miranda, A. "Structure-Activity Relationship Studies of Gomesin: Importance of the Disulfide Bridges for Conformation, Bioactivities, and Serum Stability." Biopolymers 2006, 84 (2), 205–218.

(27) Spinella, M. J., Malik, A. B., Everitt, J., Andersen, T. T. "Design and Synthesis of a Specific Endothelin 1 Antagonist: Effects on Pulmonary Vasoconstriction." Proc. Natl. Acad. Sci. 1991, 88 (16), 7443–7446.

(28) Abraham, W. M., Ahmed, A., Cortes, A., Spinella, M. J., Malik, A. B., Andersen, T. T. "A Specific Endothelin-1 Antagonist Blocks Inhaled

Endothelin-1-Induced Bronchoconstriction in Sheep." J. Appl. Physiol. 1993, 74 (5), 2537–2542.

(29) Büllesbach, E. E., Schwabe, C. "Replacement of Disulfides by Amide Bonds in the Relaxin-like Factor (RLF/INSL3) Reveals a Role for the A11– B10 Link in Transmembrane Signaling." Biochemistry (Mosc.) 2012, 51 (20), 4198–4205.

(30) Zimmermann, S., Steding, G., Emmen, J. M., Brinkmann, A. O., Nayernia, K., Holstein, A. F., Engel, W., Adham, I. M. "Targeted Disruption of the Insl3 Gene Causes Bilateral Cryptorchidism." Mol. Endocrinol. Baltim. Md 1999, 13 (5), 681–691.

(31) Shabanpoor, F., Separovic, F., Wade, J. D. "Chapter 1: The Human Insulin Superfamily of Polypeptide Hormones." In Vitamins & Hormones, Elsevier, 2009, Vol. 80, pp 1–31.

(32) Bathgate, R. A. D., Zhang, S., Hughes, R. A., Rosengren, K. J., Wade, J. D. "The Structural Determinants of Insulin-like Peptide 3 Activity." Front. Endocrinol. 2012, 3, 11.

(33) Scott, D. J., Wilkinson, T. N., Zhang, S., Ferraro, T., Wade, J. D., Tregear, G. W., Bathgate, R. A. D. "Defining the LGR8 Residues Involved in Binding Insulin-like Peptide 3." Mol. Endocrinol. Baltim. Md 2007, 21 (7), 1699–1712.

(34) Hossain, M. A., Rosengren, K. J., Haugaard-Jonsson, L. M., Zhang, S., Layfield, S., Ferraro, T., Daly, N. L., Tregear, G. W., Wade, J. D., Bathgate, R. A. D. "The A-Chain of Human Relaxin Family Peptides Has Distinct Roles in the Binding and Activation of the Different Relaxin Family Peptide Receptors." J. Biol. Chem. 2008, 283 (25), 17287–17297.

(35) 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 (19), 13068–13074.

(36) Shabanpoor, F., Hughes, R. A., Bathgate, R. A. D., Zhang, S., Scanlon, D. B., Lin, F., Hossain, M. A., Separovic, F., Wade, J. D. "Solid-Phase Synthesis of Europium-Labeled Human INSL3 as a Novel Probe for the Study of Ligand-Receptor Interactions." Bioconjug. Chem. 2008, 19 (7), 1456–1463.

(37) Zhang, S., Lin, F., Hossain, M., Shabanpoor, F., Tregear, G., Wade, J. "Simultaneous Post-cysteine(S-Acm) Group Removal Quenching of Iodine and Isolation of Peptide by One Step Ether Precipitation." Int. J. Pept. Res. Ther. 2008, 14 (4), 301–305.

(38) Najbar, L. V., Craik, D. J., Wade, J. D., Salvatore, D., McLeish, M. J. "Conformational Analysis of LYS(11-36), a Peptide Derived from the Beta- Sheet Region of T4 Lysozyme, in TFE and SDS." Biochemistry (Mosc.) 1997, 36 (38), 11525–11533.

(39) Rosengren, K. J., Lin, F., Bathgate, R. A. D., Tregear, G. W., Daly, N. L., Wade, J. D., Craik, D. J. "Solution Structure and Novel Insights into the Determinants of the Receptor Specificity of Human Relaxin-3." J. Biol. Chem. 2006, 281 (9), 5845–5851.

(40) Mountford, S. J., Liu, M., Zhang, L., Groenen, M., Herzog, H., Holliday, N. D., Thompson, P. E. "Synthetic routes to the Neuropeptide Y Y1 receptor antagonist 1229U91 and related analogues for SAR studies and cell- based imaging" Org.Biomol. Chem. 2014, 12, 3271–3281.

(41) Tickler, A. K., Clippingdale, A. B., Wade, J. D. "Amyloid-β as a “Difficult Sequence” in Solid Phase Peptide Synthesis." Protein Pept. Lett. 2004, 11 (4), 377–384.

Chapter 4

Incorporation of Cystathionine in Insulin

4.1 Introduction

Chemical incorporation of redox-stable thioether bridges in peptides has generated much interest in recent times. Strategies for the synthesis of lantibiotics, which are a class of antimicrobial peptides that are cross-linked with one or more lanthionine bridges (Figure 4.1(i)) have been studied extensively.1 Cystathionine has also attracted attention since it is isosterically similar to cystine and hence more relevant to the aims of this work (Figure 4.1(ii)). Nevertheless, it is important to consider synthetic methodologies of incorporating both thioether constructs into peptides since the strategies are mostly analogous. There are a number of approaches available to chemists for making these challenging constructs, and they can be summarized as follows: (1) incorporation of pre-formed thioether building blocks with orthogonal protection; (2) desulfurization of a disulfide bond; (3) formation of the thioether bridge on the solid support; (4) formation of the thioether bridge in solution, and (5) S-alkylation with a halogenated building block on the solid support.

Synthesis of orthogonally protected dipeptides which contain a pre-formed thioether moiety has been a widely used approach since it avoids performing chemical steps in the presence of a halogenated intermediate or other reactive species during peptide assembly. It typically requires one residue to be solid phase peptide synthesis (SPPS)-compatible and another to have one or two levels of orthogonality blocking the α-amino and α-carboxylate moieties. The ring is closed after further SPPS and lactamization (Figure 4.2(i)). Much of this pioneering work was developed by Jošt and Rudinger who described the synthesis of a cystathionine analogue suitable for tert.-butyloxycarbonyl/benzyl 2 (Boc/Bzl) chemistry. This construct was formed via an SN2 reaction with γ-halo substituted homoalanine and cysteine monomers. They were able to use this to

72 assemble an A6 – A11 cystathionine analogue of sheep insulin in order to determine whether there was any redox chemistry occurring at the site of action of the hormone.3 However, they were unable to generate a highly pure sample. Cystathionine analogues of enterotoxin have also been made in a similar fashion.4 These useful building blocks have been synthesized by a number of methods for Boc chemistry applications. Examples include reacting cysteine analogues with suitably protected 2-aziridine carboxylic acids,5 serine β- lactones6 and dehydroalanine7 building blocks for lanthionine derivatives, although the latter method requires separation of diastereoisomers.8 Fmoc- compatible constructs have also been developed in recent times. Vederas et al. have prepared a number of lantibiotic peptides via the use of building blocks with alloc/allyl protection, which provides the orthogonality to Fmoc/tBu SPPS.9,10 This same approach has been used to incorporate one11 and two12 cystathionine bridges in peptides. The latter strategy also employed para- nitrobenzyl-based protection as an extra level of orthogonality.

The desulfurization reaction of a disulfide bond has been employed for generating suitable thioether containing building blocks and also directly in cystine-containing peptides. Jacquier and co-workers synthesized a lanthionine- based construct suitable for Boc/Bzl SPPS via sulfur extrusion with tris- (diethylamino)phosphine. Shiba et al. reported the total synthesis of the peptide antibiotic nisin, which contains five lanthionine bridges, using similar chemistry.13 However, they employed a fragment condensation approach and synthesized five protected peptide intermediates with disulfide bonds then performed the desulfurization before chain assembly and global deprotection of the lantibiotic.14 The second approach, whereby an unprotected disulfide- containing peptide is treated with base, has been studied extensively (Figure 4.2(ii)). According to Galande, Spatola and co-workers, a hydroxide ion attacks the alpha proton of one of the cysteine residues and a β-elimination reaction occurs. The sulfur atom acts as the leaving group and dehydroalanine is generated, which is susceptible to nucleophilic attack from the thiolate of the other cysteine residue via a Michael addition.15,16 This reaction is generally not stereo-selective for the synthesis of lanthionines17 but it appears to be the case

73 for cystathionine bridges which can be formed from a disulfide pairing of cysteinyl and homo-cysteinyl residues.15

Formation of the thioether moiety on the solid support can occur via nucleophilic attack of a residue with a good leaving group by a free thiolate (Figure 4.2(iii)). Some of the challenges inherent in this strategy are the difficulties of incorporating a reactive, halogenated building block into the peptide chain, and efficient deprotection of the thiol prior to thioether formation. This approach has been demonstrated to be effective for the synthesis of a cystathionine analogue of oxytocin.18 Incorporation of an Fmoc-protected γ-bromohomoalaninyl residue was achieved using HOBt/DIC activation since base-catalyzed coupling conditions result in lactone formation of the building block. S-Trt on the corresponding cysteinyl residue was removed in dilute acidic conditions, followed by treatment of the resin with a non-nucleophilic base to effect the SN2 reaction. A modification of this procedure involves the incorporation of a serine (or homoserine) residue and in situ activation of the side-chain with a halogenating reagent such as carbon tetrabromide in the presence of triphenylphosphine.19 This approach has been adapted to form two over-lapping cystathionine bridges in an α-conotoxin, using S-Trt and S-thio-tert.-butyl (S-StBu) protection.

Halogenated linear peptide precursors containing a free thiol can also be prepared and isolated. Cyclization to form the thioether moiety can then occur in aqueous media (Figure 4.2(iv)). This can be achieved via in situ halogenation of a hydroxy-containing side-chain such as homoserine. This transformation should be the final synthetic step on the solid support since halides are generally not stable to Fmoc-SPPS. However, forming thioether bridges in solution is problematic and low-yielding. Side reactions relating to the halogenated residue can occur in basic buffer such as a β-elimination, hydrolysis and subsequent lactonization, which results in the peptide cleaving into two fragments.20 In order to minimize these unwanted by-products, Alewood and co-workers pioneered the use of seleno-cysteine (Sec) as a replacement for cysteine. The selenol moiety has a lower pKa than the thiol so

75 the selenate can form in acidic media, and thus avoids the base-catalyzed side reactions outlined above. This strategy has been used to mimic both lanthionine-containing20 and cystathionine-containing21 peptides with selenoethers.

It is also possible to selectively deprotect a cysteine residue and alkylate the free thiol with a suitable building block. This approach was taken for the first reported synthesis of a thioether peptide analogue, des-amino oxytocin.22 Jošt and Rudinger assembled the thiol-containing peptide precursor in solution and incorporated an iodinated construct via SN2, which was an isosteric replacement for the N-terminal cysteine residue. Lactamization and a deprotection step resulted in the synthetic target. Goodman et al. used a similar approach but performed the reaction on the solid support.23 The cyclization was also performed in solution as a semi-protected peptide. After formation of the thioether moiety, this strategy is analogous to using the pre-formed thioether building blocks. Figure 4.2(v) outlines a methodology for this approach.

When deciding on which strategy to adopt, much consideration should be given to the optimization of synthetic yields. This is particularly important when developing new chemistry for human insulin given the difficulties in its preparation. Further, if a novel compound is developed through this work then the chemistry must be scalable. If assembly of these cystathionine analogues is not efficient then they are not feasible as APIs. It appears that employing pre- formed cystathionine building blocks is the preferred method of many researchers. Recently Liu has reported excellent yields using this approach.12 However, this also requires a multi-step organic synthesis of the thioether precursors which increases complexity. The desulfurization method has merit for applications involving the substitution of one cystine moiety. But it may not be appropriate in this case as cleavage of the disulfide bond occurs during this reaction which could be problematic for insulin (and other dimeric peptides) since scrambled oligomers might be formed. Forming the thioether bridges in solution was not considered due to the low yields, and switching to selenoethers was not desired since the long term effects from daily dosages of selenoether-

76 containing APIs are relatively unknown. It was thought that in situ alkylation reactions on the solid support were more appropriate, the details of which are described below.

For this project, the A6 – A11 cystathionine analogue of human insulin was deemed to be a suitable first target (Figure 4.3). This intramolecular disulfide bond is the most straightforward to replace since the A7 – B7 and A20 – B19 intermolecular cystine bonds will require a more complex synthetic strategy, as was found in Chapter 3. The first attempt at synthesizing an A6 – A11 cystathionine analogue of human insulin involved using a modified version of the method outlined in Figure 4.2(iii). Rather than couple the halo-amino acid directly on to the peptide chain, homoserine was incorporated in to position A6 followed by in situ chlorination of the alcohol moiety with dichlorotriphenylphosphine. This approach had some success although there was not full conversion to halide. To avoid this low yielding step, the building block Fmoc-γ-Br-hAla-OH was synthesized and coupled directly (see Section 4.2). After acylation of the peptide chain with the aforementioned residue, deprotection of the S-trityl (S-Trt) cysteine residue at position A11 with dilute TFA was performed. The base-catalyzed cyclization of the peptide was attempted, but analysis indicated that this reaction led to a mixture of products. There was evidence of hydrolysis and lactonization due to trace amounts of

77 water but also loss of the bromide which was likely to be caused by β- elimination and subsequent formation of the β,γ-olefin. The failure of the cyclization to occur for this particular peptide was most likely due to steric hindrance from the tert.-butyl ethers at positions A8 and A9 and/or an unfavorable conformation on the solid support. Reaction at elevated temperatures could promote formation of the thioether bridge but would most likely accelerate the side reactions as well.

Given the difficulties in forming the cyclic intermediate and the sensitivity of the brominated residue, an alternative approach was warranted. The method reported by both Rudinger22 and Goodman23 (whereby formation of the thioether via SN2 with a suitably protected monomer proceeds first followed by a lactamization) seemed more appropriate. The reactive halogenated species is no longer present during the difficult ring-closing step which should minimize any side reactions. In the citations above, the cysteine residue was attached to the peptide first followed by thioether formation with a halogenated building block. But since the Fmoc-protected γ-bromohomoalanine derivative could be efficiently incorporated into a peptide chain and was prepared in excellent yield, an orthogonally protected cysteine building block was synthesized instead. This would mean that nucleophilic attack will occur in the reverse direction. It was decided to use the highly acid-labile Mmt protecting group for blocking of the Nα amine of cysteine.24 This is fully orthogonal to Fmoc cleavage conditions and semi-orthogonal to standard TFA labile side-chain protecting groups and the Rink amide linker. The allyl group was selected for α-carboxylate protection since it is fully orthogonal to Fmoc/tBu SPPS and the Mmt moiety. Allyl esters can be cleaved with palladium(0)25 via a palladium(0)-catalyzed allyl transfer.26 It is possible to use analogous protection for both the amino and carboxylate moieties (e.g. alloc/allyl) as described previously,11,12 but selective deprotection was proposed as it would minimize side reactions such as cross-linking on the solid support during cyclization. Therefore, Nα-mono-methoxytrityl-cysteinyl-α- allyl ester (Mmt-Cys-OAll) was synthesized to complement the brominated residue. A simple, high-yielding methodology was developed such that scale-up is feasible if required. The synthesis of both building blocks and their application

78 in the synthesis of the A6 – A11 cystathionine human insulin analogue, which will be referred to as “A11 insulin” (due to replacement of the sulfur atom with a carba moiety at position A11) are described below.

4.2 Synthesis of Fmoc-γ-Br-hAla-OH

There are two common strategies for forming halo-amino acids that are suitable for SPPS. One method involves the halogenation of an alcohol moiety using a reagent such as dichlorotriphenylphosphine. For applications such as forming thioether bridges, serine or homoserine with suitable Nα and α-carboxylate protection are typically used.23 Alternatively, aminolactones can be brominated via an acid catalyzed ring-opening reaction. The oxygen in the ester linkage is protonated and then displaced by a bromide. N-protected homoserine lactone can be treated with hydrogen bromide (HBr) in acetic acid to generate the desired γ-bromide in excellent yield.2,18 Figure 4.4 outlines an adaptation of these methods.

After solubilizing commercially available homoserine lactone hydrobromide in pyridine, an equimolar amount of 9-fluorenylmethyl N-succinimidyl carbonate (Fmoc-OSu) was added slowly in order to minimize the potential for hydrolysis of the succinimidyl ester. After four hours, the reaction mixture was added dropwise to a vigorously stirred solution of cold water whereby a white precipitate formed. After filtration, washing with water and drying at the pump, the Fmoc-homoserine lactone was redissolved in a solution of HBr in acetic acid and stirred for 18 hours. The building block was again precipitated in cold water and washed as above, followed by drying under vacuum. This synthesis of Fmoc-γ-Br-hAla-OH is simple and convenient, and the total yield was 84%. Since it does not require any refluxing or flash chromatography and, given the high yield, a scale-up synthesis is certainly feasible. The building block was fully characterized by ESI-MS, 1H and 13C NMR spectroscopy plus RP-HPLC (see Appendix 2, Section 6).

4.3 Synthesis of Mmt-Cys-OAll

In order to incorporate the necessary protecting groups into cysteine, the highly nucleophilic β-thiol (as the thiolate) should be suitably protected to avoid any side reactions. The Mmt and allyl moieties must also be stable to the cleavage conditions of the protecting group. Although the allyl ester is relatively stable, Mmt is highly sensitive to acids (by design) which means that other acid labile protecting groups for the thiol are not suitable. One possible method would be to use S-9-fluorenylmethyl (S-Fm) protection which is labile to secondary amines. Alternatively the thiol could be protected as a dimer and cleaved with a trialkylphosphine-based reducing agent. Fortunately, cystine bis-allyl ester, (H-

Cys-OAll)2, is commercially available and would significantly simplify the synthesis. A two-step approach adapted from previous protocols was employed,27 whereby the α-amino groups were alkylated with the trityl-based moiety followed by a reductive cleavage, resulting in the desired building block (Figure 4.5).

Dissolution of the (H-Cys-OAll)2 tosyl salt in THF was achieved after the addition of 5 equivalents of DIEA, presumably because the protonated species was insoluble. After the slow addition of two equivalents of mono-methoxytrityl chloride (Mmt-Cl), the reaction was allowed to stir for 4 hours. Once the alkylation was complete, 2 equivalents of tri-n-butylphosphine was added followed by 10 equivalents of water in order to effect the reduction.28 After a further 18 hours, the reaction mixture was diluted with ethyl acetate and washed

80 with brine in order to remove the oxidized phosphine and water. An acidic aqueous wash would normally be employed in this case to remove the base. However, this might cleave the sensitive Mmt moiety and, therefore, was not pursued. A purification step to remove excess phosphine was required in any case. After workup, the desired compound was eluted from a silica column with approximately 5% ethyl acetate in n-hexane. After reducing the solvent in vacuo, a colourless oil was recovered with a yield of 80%. Although a purification step was necessary, this was still a relatively simple preparation since it was a two-step one-pot synthesis and resulted in an excellent yield. RP- HPLC, ESI-MS, plus 1H and 13C NMR spectroscopy were used to characterize the building block (see Appendix 2, Section 7).

4.4 Synthesis of Human Insulin A-chain

It is important to consider the physico-chemical properties of the 21 residue A- chain when designing a synthetic route for assembling the cystathionine- bridged “A11 insulin” analogue. This peptide is prone to aggregation in solution which can make purification difficult and lead to poor yields (unpublished data). Further, it can make analysis particularly cumbersome and complex which is undesirable when developing new synthetic protocols. This is exacerbated when incorporating hydrophobic protecting groups for the cysteine side-chain, which are necessary for regio-selective formation of each cystine bond in

81 insulin. Chapter 2 outlines an enhanced synthesis using a combination of S-oNv and S-tBu as thiol protecting groups, which was used in the synthesis of native insulin.29 This strategy achieved enhanced yields but involved the use of hydrophobic protecting groups. Therefore, a pair of S-Acm protected cysteines (one for residue A7 and one for residue B7) was initially employed instead since this moiety is more hydrophilic and hence simplifies analysis and purification.

When preparing insulin-like peptides via a chemical synthesis, it is typical to isolate the A-chain intermediate functionalized with the S-Pyr moiety as well as an orthogonal protecting group such as S-Acm after the intramolecular disulfide bond has been formed. S-Pyr is an excellent leaving group during thiolysis due to its low pKa30 whereby its free thiolate “partner” on the B-chain attacks to form the A-chain/B-chain hetero-dimer. However, functionalizing the aggregation- prone A-chain with S-Pyr further increases the hydrophobicity which complicates isolation and also increases handling. But it is not possible to combine the two chains in the reverse manner (S-Pyr on the B-chain) because any attempt to isolate the A-chain with a free thiol will result in disulfide shuffling with the existing intramolecular cystine bond. Conveniently, however, this limitation does not apply in the case of isolating insulin A-chain with an A6 – A11 cystathionine bridge, since this molecular construct is non-reactive to thiols. Therefore, it will be simpler to isolate the aggregation-prone A-chain with S-Acm protection at residue A7, a free thiol at residue A20 and the A6 – A11 cystathionine bridge (Figure 4.6).

Most peptides with C-terminal carboxylates such as insulin A-chain (and B- chain) are usually assembled on a solid support with a Wang linker. However,

82 epimerization of the first amino acid can occur during loading since stronger bases such as dimethylaminopyridine (DMAP) are required to form the ester linkage. This problem can be avoided in the case of insulin A-chain since the C- terminal asparagine residue can be incorporated by attachment via the β- carboxylate side chain of Nα-9-fluorenylmethyloxycarbonyl-aspartyl-α-tert.-butyl ester (Fmoc-Asp-OtBu) to a Rink linker. This avoids the risk of epimerization during activation of the α-carboxylate31 and will be cleaved as the amide and result in the desired C-terminal asparagine residue.

Figure 4.7 outlines the synthetic strategy used for the A-chain cystathionine intermediate. Analytical data of each compound is given in Figure 4.8. The loading of the Rink linker used was 0.35 mmol/g rather than the usual 0.6 – 0.7 mmol/g in order to minimize steric crowding on the solid support. Note that S- diphenylmethyl (S-Dpm) protection32 was used at position A20, since cleavage of the N-Mmt moiety at position A6 is not chemoselective in the presence of S- Trt, which is normally used. S-Dpm is more acid-stable and thus avoids any de- blocking at position A20 and the possibility of branched products forming. After coupling of the first ten residues (I) via base-catalyzed activation protocols and microwave-assisted SPPS, Fmoc-γ-Br-hAla-OH was incorporated into the peptide chain with HOBt and DIC activation to give II. These alternative conditions were employed since the use of DIEA or a similar base can cause the bromohomoalaninyl building block to lactonize. There is also the possibility that the resin-bound nucleophilic N-terminal amino group could react with the brominated side-chain.

Mmt-Cys-OAll was then introduced in the presence of DIEA resulting in the desired thioether bridge (III) with excellent conversion (Figure 4.8(iii)). Fmoc SPPS was resumed with the incorporation of residues A7-A10 including acetamidomethyl protection for the CysA7 side-chain (IV). The peptidyl resin was then treated with the palladium(0) complex under anhydrous conditions in chloroform to cleave the allyl ester. This reaction proceeds quantitatively although at least three equivalents of palladium(0) are required. The N-terminus was then Fmoc-deprotected with piperidine as per standard protocols to give V.

Initial attempts to acylate the α-amine of the A7 residue with the α-carboxylate of the A11 residue involved the use of 1-[bis(dimethylamino)methylene]-1H- 1,2,3-triazolo[4,5-b] pyridinium 3-oxid hexafluorophosphate (HATU) in the presence of DIEA. Unfortunately the only major species that was detected had a molecular mass 98 Da more than the starting material. This indicates that the N-terminus of the peptide had most likely been capped with tetramethylguanidine33,34 which can occur when formation of the active ester is slow. This is probably because the free carboxylate of CysA6 was sterically hindered by the N-terminal residues of the peptide chain, causing this undesired side reaction to be more favourable. A second attempt was made whereby a threonine-serine pseudo-proline dipeptide35 at positions A8-A9 was incorporated in to the peptide chain. This construct mimics a proline residue and can induce a cis conformation which can aid in the cyclization.36 Unfortunately a similar outcome occurred which supports the proposition that the activation step may be the rate limiting one. A threonine-serine iso-acyl dipeptide37 was also incorporated in order to see if this changed the conformation of the resin-bound protected peptide in order to favour the lactamization. This also would have the added advantage of isolating a more water-soluble depsi-peptide due to the auxiliary amino group. This tactic also failed so a new approach was taken whereby the guanidinyl side product could not form. HOBt/DIC chemistry was used instead (with conventional building blocks on the peptidyl resin) and after a single treatment with five equivalents of both reagents each the reaction was approximately one third complete. Two subsequent treatments pushed the reaction to completion to give VI. Although the kinetics were slow, the RP- HPLC profile was relatively clean (Figure 4.8(vi)) which is the critical factor. No detectable cross-linking was observed.

Mmt was then removed with a dilute TFA solution in dichloromethane in order to de-block the N-terminus. Multiple treatments of the acidic solution were required since the trityl cations formed are in equilibrium with the protected moiety. The use of additives such as TIPS, which is an excellent scavenger, can speed up this deprotection. The remaining five residues were added to extend the peptide

86 chain to completion (VII). Cleavage from the solid support and global deprotection of the amino acid side-chains was performed using a cocktail of

TFA, TIPS, H2O and 3,6-dioxa-1,8-octane-dithiol (DODT). After ether precipitation and decanting, the resulting white powder was dried. RP-HPLC and MALDI-MS analysis confirmed that the single major product was the desired cyclized intermediate (VIII).

As stated above, purification of various analogues of the insulin A-chain is problematic due to its propensity to self-aggregate in solution. To avoid this, the peptide was dissolved and quickly filtered and injected onto the column to minimize the occurrence of this oligomerization. Standard aqueous TFA buffers and a linear gradient of acetonitrile were used. Once the product eluted from the column, the solution was quickly frozen with liquid nitrogen in order to minimize any further aggregation. The purified fractions were lyophilized, resulting in a yield of 12% as calculated from crude A-chain which is an excellent result considering the complexity of the synthesis. The construct was characterized by RP-HPLC and MALDI-MS which indicated a purity of ~80%. Under normal circumstances, a second purification might have been advisable but it was thought that any further handling in solution could significantly compromise yields.

4.5 Synthesis of B-chain

The assembly of the B-chain is more straightforward given that it is a simple linear sequence. Standard amino acid building blocks can be used for the synthesis and a low-loading resin was used (0.18 mmol/g) in order to maximize the quality of the crude material. Microwave-assisted automated SPPS was again employed with low power used during the coupling of cysteine and histidine building blocks in order to minimize any epimerization that may occur. Cys(Acm) was incorporated at position B7 since it must be paired with residue A7. Acid-labile S-Trt protection was used at B19 in order to generate the free thiol post-cleavage (Figure 4.9). The proposed reaction scheme is outlined in

Figure 4.10 and the analytical data associated with this synthesis is depicted in Figure 4.11. The cleavage and deprotection protocol employed was the same as that used for the A-chain. The product recovered had a relatively high purity for a crude peptide. It is typical to purify most peptides at pH 2 with TFA-based buffers. This causes basic residues (Arg, His, Lys) and the N-terminus to be cationic and thus increases the solubility of the peptide. Unfortunately various B-chain analogues of insulin tend to elute from reversed-phase HPLC columns as a broad peak which is obviously undesirable since it makes purification more difficult. It is likely that the cationic B-chain adopts a number of possible conformations in solution, some of which may have slightly altered retention times.

Therefore, an alternative approach was undertaken whereby the behaviour of the B-chain was observed at a higher pH value. This strategy has been explored at length with amyloid-beta peptides which are notoriously difficult to purify due to their propensity for aggregation.38 It was proposed that performing RP-HPLC analysis and purification at pH 9 using 10 mM ammonium acetate buffers, which would deprotonate acidic residues such as aspartate and, in the case of the insulin B-chain, the two glutamate residues and the C-terminus. This level of basicity was selected as a compromise between ensuring that the peptide was suitably anionic but not too alkaline such that de-amidation of asparagine and glutamine residues would occur. It was envisaged that making the peptide negatively charged would alter the physico-chemical properties such that a more structured conformation would be induced leading to a “sharper” chromatographic peak which was in fact observed (Figure 4.11(i)). Therefore, analysis and purification in ammonium acetate buffers was pursued. It is important to note that not all stationary phase media for RP-HPLC are stable to basic conditions so column selection is critical.

The B-chain with acetamidomethyl protection at B7 was purified in two batches. The purification profile shown in Figure 4.11(ii) indicated that separation of most of the peptidic impurities was successful and there was no detectable presence of a B-chain homo-dimer which was a possibility due to the relatively high pH.

Purging of the HPLC buffers with helium displaces any elemental oxygen and most likely suppressed this reaction. After pooling the purified fractions, the solution was lyophilized to generate IX at a yield of 16% as calculated from crude B-chain.

As outlined above, S-pyridinesulfenyl functionalization usually occurs on the A- chain when forming hetero-dimers of insulin-like peptides. The insulin B-chain elutes at a relatively longer retention time and, therefore, it is not known what effect incorporating another hydrophobic moiety will have on its behaviour during analysis and purification. Nevertheless, the same chromatographic conditions were used as previously given our earlier success. The reaction was performed in neat TFA with dipyridine disulfide (DPDS). A large excess of reagent was used in order to favor the functionalized monomer over the B-chain dimer. Using TFA as the solvent also minimized dimerization because the free cysteine side-chain was mostly protonated and therefore unreactive, although this did slow the rate of S-Pyr functionalization. Figure 4.11(iii) highlights the reaction as monitored on an analytical C18 RP-HPLC column. The purification was straightforward despite the longer retention time on the preparative column (Figure 4.11(iv)). The B-chain precursor X was isolated as a highly pure intermediate with a yield of 33% (Figures 4.11(v) and (vi)). The total yield for the B-chain intermediate was 5%.

4.6 Combination of A-chain and B-chain

Formation of the insulin hetero-dimer via asymmetrical combination of A- and B- chain intermediates was then undertaken (Figure 4.12). The key feature of this chemistry is that one construct has a free thiol and the other has an activated thiol. The latter is typically functionalized with a leaving group (such as S-Pyr) which should ideally rearrange to form a non-reactive adduct.39 The alternative to this is to employ a symmetrical strategy whereby the two free thiols of each construct are oxidized, but this approach leads to the formation of homo-dimers plus other higher oligomers and results in diminished yields. However, with the

90 former method, careful optimization of the pH and buffer composition can result in much higher yields. Therefore, this strategy has found widespread use and especially for insulin-like peptides.40

Particular care must be taken when handling the insulin A-chain in solution due to its propensity to aggregate but this risk is partly offset by the use of a strong denaturant such as 6 M guanidinium hydrochloride. The rate of reaction is heavily dependent on the basicity of the solution. Elevating the pH can shift the thiol/thiolate equilibrium towards the reactive nucleophilic species, which will accelerate thiolysis. However, employing these conditions will also make it more likely that formation of the A-chain homo-dimer will occur, particularly if the combination buffer has not been degassed. Therefore, determining the optimal pH for this particular system is a trade-off between a faster reaction time to minimize any A-chain aggregation and avoiding any A-chain homo-dimer being formed. The reaction was performed at pH 8 for 90 minutes with equimolar amounts of each monomer and monitored via RP-HPLC in ammonium acetate buffers. As expected, the hetero-dimeric species XI was the major product and eluted in between the two monomeric precursors (Figure 4.13(i)). There was no A-chain homo-dimer detected but interestingly some B-chain homo-dimer was formed. This indicates that some disulfide shuffling occurred. The synthetic insulin intermediate was purified at pH 9 using standard protocols, with a yield of 45%.

4.7 Formation of the A7-B7 Disulfide Bond

Formation of disulfide bonds in the presence of existing ones has always been challenging and its success is often dependent on the sequence and also the order in which each is formed. In this case, a symmetrical strategy using a pair of S-Acm protecting groups was pursued (Figure 4.14) which is used routinely in the formation of disulfide bonds in insulin-like peptides.31,40,41 Essentially, reaction occurs via an oxidative cleavage using an excess of iodine. Unfortunately, yields are typically low due to modification of sensitive residues such as methionine, tryptophan and tyrosine, and also disulfide shuffling. Fortunately, insulin only contains tyrosine and the thioether bridge which can also be oxidized, so it is possible that side reactions will be minimal. There are only two cystine bonds in this particular analogue so disulfide scrambling should also be less likely.

A number of solvent systems can be used when performing this reaction. Acetic acid with a dilute solution of hydrochloric acid as an additive was selected since this mixture is used routinely for unprotected peptides.40 A dilute peptide concentration is necessary in this instance as an intra-molecular reaction is desired. If the concentration is too high then formation of an inter-molecular disulfide bond may be more favoured. The optimal quantity of iodine to be used can vary from peptide to peptide and is a balance between minimizing oxidation of sensitive residues and driving the reaction to completion. In this instance 20 equivalents was used for 45 minutes and resulted in a clean reaction with one primary discrete product forming that eluted about one minute earlier on the HPLC column (Figure 4.15(i)). Approximately 80% of the starting material had disappeared but there was difficulty in converting the remainder to the final product XII. Nevertheless, a relatively high yield was realized for this difficult reaction. The excess iodine was quenched with a solution of ascorbic acid which turned the solution from brown to colourless. The acetic acid solution was diluted by half such that it could be loaded directly on to a HPLC column without premature elution. This strategy avoids any potential losses of material which can occur during work-ups such as ether precipitations that are sometimes used.

Ammonium acetate buffers were again used for the initial purification, which resulted in a purity of 95%. Using these buffers generates peptides as ammonium salts which are paired with acidic residues and the C-termini. It was decided that this counter-ion should be exchanged with TFA since other peptides have been characterized and tested in this form. Further, performing a second purification with different buffers altered the selectivity of impurities that were co-eluting with the parent peptide and thus resulted in a more highly pure product. The yield of the iodine oxidation was 25% which is excellent for this type of reaction. This may be because there were only two disulfide bonds present which limits the number of permutations possible with respect to disulfide connectivity. The overall folding yield of the synthesis as calculated from purified A-chain was found to be 11%.

4.8 Conclusion

The work reported in this chapter has shown that it is feasible to replace the A6 – A11 cystine moiety in human insulin with a non-reducible thioether bridge. After careful evaluation of the various approaches for incorporating cystathionine bridges in peptides, a simple and relatively high yielding method was adopted. This involved the use of two “unusual” building blocks: a halo- amino acid and an orthogonally protected cysteine derivative. After

94 incorporation of the halo-amino acid on to the peptide chain, the thioether moiety was formed via an SN2 reaction with the nucleophilic β-thiol group of cysteine. Further peptide synthesis and cleavage of the orthogonal protecting groups was then performed, followed by a lactamization as the cyclization step.

The yields realized during this difficult synthesis were comparable to that of other peptides of similar complexity. This is largely due to the simplicity and efficiency of the chemistry but also because of optimized purification protocols in basic as well as acidic buffers. The syntheses of Fmoc-γ-Br-hAla-OH and Mmt-Cys-OAll were also very straightforward and high-yielding. This is important since any lead compounds developed from this work should be scalable. There is, of course, scope to further improve the recovery of the final product once each chemical step is carefully optimized. The final purity of the A11 insulin analogue was 96% as assessed by RP-HPLC and MALDI-TOF MS. This can be partly attributable to the effective chemistry used, which led to fewer impurities and, therefore, fewer intractable purification issues. The methodologies developed for this synthesis can form the basis for assembling other cystathionine analogues of insulin, namely replacement of the A20 – B19 and A7 – B7 disulfide bonds. They can also be applied to other therapeutically relevant disulfide rich peptides such as conotoxins.

4.9 References

(1) Chatterjee, C., Paul, M., Xie, L., van der Donk, W. A. "Biosynthesis and Mode of Action of Lantibiotics." Chem. Rev. 2005, 105 (2), 633–684.

(2) Jošt, K., Rudinger, J. "Amino Acids and Peptides LXXIV. Derivatives of L- Cystathionine Suitable for Peptide Synthesis." 1967, 32, 2485–2490.

(3) Jošt, K., Rudinger, J., Klostermeyer, H., Zahn, H. "Synthese Und Hypoglycämische Wirkung Eines Insulinanalogen Cystathionin-Peptides: Ein Argument Gegen Die Beteiligung Der Intrachenaren Disulfidgruppe Bei Der Insulin Wirkung." Z. Für Naturforschung B 1968, 23 b, 1059– 1061.

(4) Hidaka, Y., Ohmori, K., Wada, A., Ozaki, H., Ito, H., Hirayama, T., Takeda, Y., Shimonishi, Y. "Synthesis and Biological Properties of Carba- Analogs of Heat-Stable Enterotoxin (ST) Produced by Enterotoxigenic Escherichia Coli." Biochem. Biophys. Res. Commun. 1991, 176 (3), 958– 965.

(5) Nakajima, K., Oda, H., Okawa, K. "Studies on 2-Aziridinecarboxylic Acid. IX Convenient Synthesis of Optically Active S-Alkylcysteine, Threo-S- Alkyl-B-Methylcysteine, and Lanthionine Derivatives via the Ring-Opening Reaction of Aziridine by Several Thiols." Bull. Chem. Soc. Jpn. 1983, 56 (2), 520–522.

(6) Shao, H., Wang, S. H., Lee, C.-W., Oesapay, G., Goodman, M. A Facile "Synthesis of Orthogonally Protected Stereoisomeric Lanthionines by Regioselective Ring Opening of Serine. Beta.-Lactone Derivatives." J. Org. Chem. 1995, 60 (10), 2956–2957.

(7) Ösapay, G., Goodman, M. "New Application of Peptide Cyclization on an Oxime Resin (the PCOR Method): Preparation of Lanthionine Peptides." J. Chem. Soc. Chem. Commun. 1993, No. 21, 1599–1600.

(8) Probert, J. M., Rennex, D., Bradley, M. "Lanthionines for Solid Phase Synthesis." Tetrahedron Lett. 1996, 37 (7), 1101–1104.

(9) Pattabiraman, V. R., McKinnie, S. M. K., Vederas, J. C. "Solid-Supported Synthesis and Biological Evaluation of the Lantibiotic Peptide Bis(desmethyl) Lacticin 3147 A2." Angew. Chem. Int. Ed. 2008, 47, 9472– 9475.

(10) Ross, A. C., Liu, H., Pattabiraman, V. R., Vederas, J. C. "Synthesis of the Lantibiotic Lactocin S Using Peptide Cyclizations on Solid Phase." J. Am. Chem. Soc. 2010, 132, 462–463.

(11) Knerr, P. J., Tzekou, A., Ricklin, D., Qu, H., Chen, H., van der Donk, W. A., Lambris, J. D. "Synthesis and Activity of Thioether-Containing Analogues of the Complement Inhibitor Compstatin." Chem. Biol. 2011, 6, 753–760.

(12) Cui, H.-K., Guo, Y., He, Y., Wang, F.-L., Chang, H.-N., Wang, Y.-J., Wu, F.-M., Tian, C.-L., Liu, L. "Diaminodiacid-Based Solid-Phase Synthesis of Peptide Disulfide Bond Mimics." Angew. Chem. Int. Ed. 2013, 52, 1–6.

(13) Fukase, K., Kitazawa, M., Sano, A., Shimbo, K., Horimoto, S., Fujita, H., Kubo, A., Wakamiya, T., Shiba, T. "Synthetic Study on Peptide Antibiotic Nisin. V. Total Synthesis of Nisin." Bull. Chem. Soc. Jpn. 1992, 65, 2227– 2240.

(14) Fukase, K., Oda, Y., Kubo, A., Wakamiya, T., Shiba, T. "Synthetic Study on Peptide Antibiotic Nisin. IV. Synthesis of Ring D-E." Bull. Chem. Soc. Jpn. 1990, 63, 1758–1763.

(15) Galande, A. K., Trent, J. O., Spatola, A. F. "Understanding Base-Assisted Desulfurization Using a Variety of Disulfide-Bridged Peptides." Pept. Sci. 2003, 71 (5), 534–551.

(16) Galande, A. K., Bramlett, K. S., Burris, T. P., Wittliff, J. L., Spatola, A. F. "Thioether Side Chain Cyclization for Helical Peptide Formation: Inhibitors

of Estrogen Receptor–coactivator Interactions." J. Pept. Res. 2004, 63 (3), 297–302.

(17) Pulka-Ziach, K., Pavet, V., Chekkat, N., Estieu-Gionnet, K., Rohac, R., Lechner, M.-C., Smulski, C. R., Zeder-Lutz, G., Altschuh, D., Gronemeyer, H., Fournel, S., Odaert, B., Guichard, G. "Thioether Analogues of Disulfide-Bridged Cyclic Peptides Targeting Death Receptor 5: Conformational Analysis, Dimerisation and Consequences for Receptor Activation." ChemBioChem 2015, 16, 293–301.

(18) Mayer, J. P., Heil, J. R., Zhang, J., Munson, M. C. "An Alternative Solid- Phase Approach to C 1 -Oxytocin." Tetrahedron Lett. 1995, 36 (41), 7387–7390.

(19) Mayer, J. P., Zhang, J., Groeger, S., LIU, C.-F., Jarosinski, M. A. "Lanthionine Macrocyclization by in Situ Activation of Serine." J. Pept. Res. 1998, 51 (6), 432–436.

(20) Dantas de Araujo, A., Mobli, M., King, G. F., Alewood, P. F. "Cyclization of Peptides by Using Selenolanthionine Bridges." Angew. Chem. Int. Ed. 2012, 124, 10444–10448.

(21) Dantas de Araujo, A., Mobli, M., Castro, J., Harrington, A. M., Vetter, I., Dekan, Z., Muttenhaler, M., Wan, J., Lewis, R. J., King, G. F., Brierley, S. M., Alewood, P. F. "Selenoether Oxytocin Analogues Have Analgesic Properties in a Mouse Model of Chronic Abdominal Pain." Nat. Commun. 2014, 5, 3165.

(22) Rudinger, J., Jošt, K. "A Biologically Active Analogue of Oxytocin Not Containing a Disulfide Group." Experientia 1964, 20, 570–571.

(23) Rew, Y., Malkmus, S., Svensson, C., Yaksh, T. L., Chung, N. N., Schiller, P. W., Cassel, J. A., DeHaven, R. N., Goodman, M. "Synthesis and Biological Activities of Cyclic Lanthionine Enkephalin Analogues: δ - Opioid Receptor Selective Ligands." J. Med. Chem. 2002, 45 (17), 3746– 3754.

(24) Dubowchik, G. M., Radia, S. "Monomethoxytrityl (MMT) as a Versatile Amino Protecting Group for Complex Prodrugs of Anticancer Compounds Sensitive to Strong Acids, Bases and Nucleophiles." Tetrahedron Lett. 1997, 38 (30), 5257–5260.

(25) Kunz, H., Waldmann, H., Unverzagt, C. "Allyl Ester as Temporary Protecting Group for the β-Carboxy Function of Aspartic Acid." Int. J. Pept. Protein Res. 1985, 26 (5), 493–497.

(26) Trost, B. M. "New Rules of Selectivity: Allylic Alkylations Catalyzed by Palladium." Acc. Chem. Res. 1980, 13 (11), 385–393.

(27) Dauty, E., Remy, J.-S., Blessing, T., Behr, J.-P. "Dimerizable Cationic Detergents with a Low Cmc Condense Plasmid DNA into Nanometric Particles and Transfect Cells in Culture." J. Am. Chem. Soc. 2001, 123 (38), 9227–9234.

(28) Sweetman, B. J., Maclaren, J. A. "The Reduction of Wool Keratin by Tertiary Phosphines." Aust. J. Chem. 1966, 19 (12), 2347–2354.

(29) 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 (31), 9549–9552.

(30) Rabanal, F., DeGrado, W. F., Dutton, P. L. "Use of 2,2’- dithiobis(5nitropyridine) for the Heterodimerization of Cysteine Containing Peptides. Introduction of the 5-Nitro-2-Pyridinesulfenyl Group." Tetrahedron Lett. 1996, 37 (9), 1347–1350.

(31) 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 (24), 11384–11392.

(32) Góngora-Benítez, M., Mendive-Tapia, L., Ramos-Tomillero, I., Breman, A. C., Tulla-Puche, J., Albericio, F. "Acid-Labile Cys-Protecting Groups for the Fmoc/ t Bu Strategy: Filling the Gap." Org. Lett. 2012, 14 (21), 5472– 5475.

(33) Story, S. C., Aldrich, J. V. "Side-Product Formation during Cyclization with HBTU on a Solid Support." Int. J. Pept. Protein Res. 1994, 43, 292–296.

(34) Albericio, F., Bofill, J. M., El-Faham, A., Kates, S. A. "Use of Onium Salt- Based Coupling Reagents in Peptide Synthesis." J. Org. Chem. 1998, 63 (26), 9678–9683.

(35) Wöhr, T., Wahl, F., Nefzi, A., Rohwedder, B., Sato, T., Sun, X., Mutter, M. "Pseudo-Prolines as a Solubilizing, Structure-Disrupting Protection Technique in Peptide Synthesis." J. Am. Chem. Soc. 1996, 118 (39), 9218–9227.

(36) van Lierop, B. J., Whelan, A. N., Andrikopoulos, S., Mulder, R. J., Jackson, W. R., Robinson, A. J. "Methods for Enhancing Ring Closing Metathesis Yield in Peptides: Synthesis of a Dicarba Human Growth Hormone Fragment." Int. J. Pept. Res. Ther. 2010, 16 (3), 133–144.

(37) Yoshiya, T., Taniguchi, A., Sohma, Y., Fukao, F., Nakamura, S., Abe, N., Ito, N., Skwarczynski, M., Kimura, T., Hayashi, Y., Kiso, Y. O-Acyl Isopeptide Method? For Peptide Synthesis: Synthesis of Forty Kinds of O-Acyl Isodipeptide Unit? Boc-Ser/Thr(Fmoc-Xaa)-OH. Org. Biomol. Chem. 2007, 5 (11), 1720.

(38) Kok, W. M., Cottam, J. M., Ciccotosto, G. D., Miles, L. A., Karas, J. A., Scanlon, D. B., Roberts, B. R., Parker, M. W., Cappai, R., Barnham, K. J., Hutton, C. A. "Synthetic Dityrosine-Linked β-Amyloid Dimers Form Stable, Soluble, Neurotoxic Oligomers." Chem. Sci. 2013, 4 (12), 4449.

(39) Maruyama, K., Nagata, K., Tanaka, M., Nagasawa, H., Isogai, A., Ishizaki, H., Suzuki, A. "Synthesis of Bombyxin-IV, an Insulin Superfamily Peptide

100

from the Silkworm, Bombyx Mori, by Stepwise and Selective Formation of Three Disulfide Bridges." J. Protein Chem. 1992, 11 (1), 1–12.

(40) 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., Gundlach, A. L., Samuel, C. S., Tregear, G. W., Wade, J. D. "Relaxin-3: Improved Synthesis Strategy and Demonstration of Its High-Affinity Interaction with the Relaxin Receptor LGR7 Both In Vitro and In Vivo." Biochemistry (Mosc.) 2006, 45 (3), 1043–1053.

(41) Sieber, P., Kamber, B., Hartmann, A., Jöhl, A., Riniker, B., Rittel, W. "Totalsynthese von Humaninsulin. IV. Beschreibung Der Endstufen." Helv. Chim. Acta 1977, 60 (1), 27–37.

101

102

Chapter 5

Biological and Chemical Characterization of A11 Insulin

103

5.1 Introduction

It is essential that all new synthetic peptides, peptidomimetics and analogues are subjected to comprehensive chemical characterization to confirm their purity prior to biological and structural studies. It is now accepted that the minimum criteria for assessing purity is physico-chemical, biological and, if possible, structural analysis. Once the purity of the synthetic product has been confirmed to be of an acceptable level, then the effect of the chemical modification can be assessed and compared against ‘gold’ standards or the native hormone. Finally, more specific properties such as thermal stability can be investigated as well. This approach can be applied to the analysis of A11 insulin in order to study the effects of replacing the relevant β-sulfur atom with a methylene group. The characterization of this analogue can be performed using a variety of techniques in an attempt to determine whether it is more thermally stable than native insulin and maintains native activity, structure and metabolic stability. Firstly the binding affinity for the insulin receptor is determined. This result is critical since any significant decrease in activity may render this analogue unviable as a lead compound. The secondary structure can also be investigated by circular dichroism (CD) spectroscopy to measure α-helicity and comparisons can be made with the native hormone. Furthermore the tertiary structure can, where possible, be determined via proton nuclear magnetic resonance spectroscopy (1H NMR) and conducting two-dimensional experiments.

The metabolic stability of A11 insulin can be investigated via a human serum stability assay. This is important because even minor modifications to the cystine framework can significantly reduce the plasma half-life.1 But the most relevant property for evaluation for this project is the thermal stability of the new analogue since one of the primary aims of this work is to develop an insulin therapeutic that is more resistant to degradation at higher temperatures. Differential scanning calorimetry (DSC) is a useful tool to determine transition temperatures of conformational or covalent changes in proteins, so this technique can give insights into the relative stability of A11 insulin compared with the native hormone. The most direct way of determining thermal stability is

104 via incubation at an elevated temperature and monitoring peptide degradation via RP-HPLC. All these methods were applied to A11 insulin and the results are herein described.

5.2 Receptor Binding Activity

The insulin receptor’s (IR) function is to mediate glucose uptake into cells when activated by its endogenous ligand. It is a tyrosine kinase and consists of two extracellular α subunits and two transmembrane β subunits linked by disulfide bonds.2 There are two isoforms of the receptor, IR-A and IR-B. The former has exon 11 included which is situated at the C-terminus of the α subunit and alters the selectivity of both isoforms in terms of ligand specificity. Insulin-like growth factor 2 (IGF-II) binds to IR-A with an approximately 100-fold increase in affinity compared to IR-B.3 IR-B is mostly expressed in liver, muscle, adipose and kidney tissues and is involved in the metabolic effects of insulin. IR-A is ubiquitous in adults and also over-expressed in certain cancer cell types, so the response by these cells towards insulin and IGF-II could be different to that of normal cells.2 Any selectivity towards either IR-A or IR-B exhibited by new agonists such as A11 insulin would be interesting and could potentially be beneficial. In any case, it is desirable that there is at least the same affinity for IR-B; any reduction in activity would require a higher dosage to produce the same biological effect. Given these issues, receptor binding activity should be determined for both IR isoforms.

The experimental procedure for determining the binding affinities of ligands for both IR-A and IR-B are similar. Cells that over-express the relevant insulin receptor were lysed and centrifuged. The supernatant containing IR-A or IR-B was then transferred to wells coated with anti-IR antibody in order to “capture” the receptors. Each coated well was incubated with europium-labelled insulin (Eu-insulin) and various concentrations of ligand followed by washing of the plates. The fluorescence of each well was measured in order to determine the amount of Eu-insulin that was displaced by the ligand to be tested.4

105

Figure 5.1(i) shows the competition binding curve of Eu-insulin binding to human IR-B. A11 insulin clearly has the same affinity for this receptor as the native hormone given the tight correlation of the two binding curves. Furthermore there is no statistically significant difference between the half maximal inhibitory concentration (IC50) values, which means that A11 insulin will displace 50% of Eu-insulin at the same concentration as the native form. This result indicates that incorporation of cystathionine within the A-chain did not significantly alter the tertiary structure of the new analogue to adversely affect its binding to IR-B.

Figure 5.1(ii) shows the competition binding curve of Eu-insulin binding to human IR-A. The data correlates relatively closely with native insulin and, again, the IC50 values are similar which indicated that replacement of the β-sulfur at A11 with a methylene group did not significantly change the affinity the analogue has for IR-A. It can be concluded A11 insulin behaves in much the same way as the native hormone in terms of binding affinity for both receptor isoforms. Both assays were performed by Assoc. Prof. Briony Forbes.

106

5.3 Circular Dichroism Spectroscopy

Given the positive results in the binding assays, it is reasonable to expect that A11 insulin would have similar secondary structural characteristics as that of native insulin. This can be determined by CD spectroscopy which uses circularly polarized light to probe optically active proteins such as insulin. Figure 5.2 shows the superimposed CD spectra of both the A11 analogue and native insulin in water at 25 oC. The large positive peak at 195 – 200 nm indicates a significant proportion of α-helicity in both proteins, as does the minima at 208 and 220 nm. The mean residual weight ellipticity (MRE) at 222 nm can be used to approximate the proportion of α-helicity.5 In water, it was found that native insulin had α-helical content of ~24% ([θ]222 = -9,065) and A11 insulin had ~25%

([θ]222 = -9,406). The proportion of α-helicity is smaller than what has been determined by X-ray crystallography studies, where it was found that there are three α-helical segments in insulin: A1 – A6, A16 – A21 and B10 – B19 or approximately 43% of the molecule.6

107

Given the close similarity of both spectra, it is clear that there has been little perturbation of the secondary structure from the incorporation of the cystathionine bridge and largely explains why binding activity was maintained. This result compares more favourably to the CD spectra of A10 – A15 unsaturated di-carba analogues of human INSL37 and human relaxin-38 (although one isomer does correlate closely with the latter). This anomaly can be attributed to the fact that a thioether bridge is more isosterically similar to a disulfide bond than an unsaturated di-carba bond. Analogues of insulin with triazole bridges replacing the A7 – B7 disulfide bond have also been analyzed by CD spectroscopy and the data indicated that there was significant disruption of the α-helical content.9 This finding correlates with the fact that binding affinity for the insulin receptor was significantly decreased and is not surprising given the marked structural and rotational differences of the triazole moiety compared to the disulfide. This provides further support for the use of cystathionine as a replacement cystine mimic. These experiments were conducted under the guidance of Dr Marco Sani.

5.4 1H Nuclear Magnetic Resonance Spectroscopy

To more precisely determine the three-dimensional shape of the A11 insulin analogue, 1H NMR spectroscopy can be utilized since CD spectroscopy provides only limited structural information. One of the difficulties in working with insulin is that it has the propensity to dimerize.10 Some solution-phase NMR experiments on insulin have used mutated analogues which have the binding at the dimer interface disrupted11 or used solutions with pH values at below 2, whereby the monomeric form is more favoured.12 Nevertheless initial experiments can be performed at higher pH values in order to make comparisons between A11 insulin and the native hormone. Preliminary data for two-dimensional experiments13 were collected for structure elucidation purposes; the one-dimensional spectrum was also obtained to make comparisons between both peptides.

108

250 µM deuterated aqueous solutions of both the cystathionine analogue and native insulin were made up and analyzed at 25 oC using an 800 MHz NMR spectrometer. Examination of the one-dimensional spectra of both proteins revealed a high level of similarity between the two. The N-H region is shown in

Figures 5.3(i) and 5.3(ii), and the Hα region is shown in Figures 5.3(iii) and 5.3(iv). These spectra suggest that the tertiary structure of A11 insulin closely mimics that of the native. But it is important to note that a number of peaks in the Hα region of the spectrum have slightly altered chemical shifts. This would indicate that the shorter carbon–sulfur bond length in the cystathionine bridge has caused a slight perturbation in the three-dimensional structure of the insulin molecule but not enough to significantly alter the secondary structure or the binding affinity for IR-A and IR-B. Although it is difficult to draw too many conclusions from this set of data without properly assigning each peak, it does show well-resolved signals which indicate that A11 insulin is a highly ordered molecule. This should allow expedient determination of the solution-phase structure under the conditions stated above. If significant dimerization is occurring and complicating the analysis, then alternative conditions can be employed such as performing dilutions or altering the pH such that dimer formation is unfavorable. Dr Marco Sani assisted in these experiments.

109

5.5 Differential Scanning Calorimetry

A key objective of this project is to develop insulin analogues with enhanced thermal stability. One way of understanding the thermodynamic properties of a protein is to employ techniques such as DSC. This powerful technique can detect phase transitions in biomacromolecules by subjecting a given sample and a reference to a temperature gradient. It then detects exothermic (or endothermic) events in the sample by measuring the energy required to equalize the temperature of the reference.14 This method has already been used to study the thermal stability of the insulin hexamer15 and a monomeric analogue.16 Therefore, DSC is a useful method to gain insights into the thermodynamic stability of A11 insulin as compared to the native form.

Figures 5.4(i) and 5.4(ii) show the DSC curves for both the cystathionine analogue and native insulin, respectively. There are some clear similarities between both profiles in that a major exothermic phase transition occurs and is

110 followed by a second minor one at approximately 7 oC higher. It should be noted that each DSC curve was accurately reproduced after the thermal gradient was repeated. This indicates that these phase transitions are reversible and most likely reflect conformational changes of the insulin molecules rather than covalent ones. Interestingly, the A11 analogue appears to undergo a transition change at a lower temperature than the native hormone. The altered geometry in the A6 – A11 bridge may slightly destabilize its tertiary structure which could account for this anomaly. Alternatively, the initial phase transition may represent dissociation of oligomeric species given the possibility of an equilibrium between monomeric and dimeric insulin in both samples. Therefore, the A-chain modification may be destabilizing the dimer interface. More DSC experiments are required in order to further understand the nature of the phase transitions for both analogues. Dr Marco Sani provided assistance for these experiments.

5.6 Human Serum Stability Assay

It is critical that plasma half life is maintained for the A11 analogue otherwise it will not be deemed suitable as a potential therapeutic. A recent manuscript on E and Z unsaturated di-carba analogues of H2 relaxin reported that these modifications caused a significant decrease in serum stability in vitro, from 7 hours to 1 hour.1 This was most likely due to conformational changes which further expose amide bonds that are sensitive to proteolysis given the subtle differences in the CD spectra. The half life of insulin in vivo is quite short and is estimated to be 4-6 minutes.17 This is mainly due to rapid clearance by the liver and kidneys as well as intracellular metabolism via insulin-degrading enzyme and other proteases.18 Minimal proteolysis actually occurs in circulation although some degradation can take place in erythrocytes via receptor- mediated uptake of the hormone.19 But insulin is reasonably stable in plasma in vitro, and it has been found that minimal degradation occurs after 1 week at 4 oC. At physiological temperature, i.e., 37 oC, the half life is approximately 2-3 days.20

111

An experiment was undertaken to determine whether the A-chain cystathionine bridge affects the serum stability of insulin. Both the native hormone and the A11 analogue were both incubated in 100% human plasma at 37 oC and aliquots were removed at specific time points, precipitated with an acetonitrile- based solution and analyzed via RP-HPLC as per a previously reported procedure21 (see Appendix 2, Section 13). Figure 5.5 shows the degradation profiles of both analogues over 8 hours, and there appears to be little difference between the two. But more importantly, these preliminary results show that there is no significant decrease in serum stability which is in direct contrast to the data obtained for the di-carba analogues of H2 relaxin mentioned above. Given that the CD and 1H NMR spectra of A11 insulin closely mimics that of the native form and that it exhibits native binding activity, it is likely that there are no subtle conformational changes which expose sensitive amide bonds to proteases. This is perhaps not such a surprising result and further demonstrates that cystathionine is an excellent mimic of cystine.

112

5.7 Thermal Stability Assay

Another way to determine the thermal stability of insulin and its analogues is to subject each sample to direct heat stress and measure the rate of degradation. This can be achieved by analyzing sample aliquots at specific time points via techniques such as RP-HPLC whereby the area under the peak is calculated. Most insulin studies such as these have used 37 oC as a basis for measuring thermal stability because of the relevance to insulin pumps.15,22 In this instance physiological temperature is not applicable because the objective is to develop heat stable analogues that can be more easily stored without refrigeration.

The temperature chosen for the A11 insulin thermal stability assay was 60 oC. This value is somewhat arbitrary but lower temperatures would require extended incubation times. Phillips and Weiss observed only modest degradation of a monomeric insulin analogue at 37 oC over three months. Furthermore, applying higher temperatures is not without precedent. The thermal degradation of an oxytocin analogue with a selenoether bond replacing the disulfide was measured at 55 oC.23 Therefore, 60 oC is a practical starting point for this particular assay in order to generate preliminary data in a realistic window of time.

Both A11 and native insulin were dissolved in water and incubated for four days in triplicate, followed by analysis on a C18 RP-HPLC column (see Appendix 2). For native insulin it is apparent that the corresponding peak significantly diminished over the time course but the peak height for the A11 analogue appears not to have significantly changed. This is confirmed by plotting the percentage of peptide degradation versus time (Figure 5.6). After four days less than 40% of the native peptide remained compared to approximately 90% for A11. Although this data is still preliminary, a clear difference in thermal stability between the two was established.

The degradation products will most likely be higher order aggregates. With respect to the disulfide bonds, it is possible that the pathway to covalent oligomers is blocked by the incorporation of the non-reducible thioether bridge.

113

Although A7 – B7 cystine is implicated in cleavage and thiolysis, it is possible that the A6 residue is also involved and the stable mimetic has hindered this process. This result may initially appear contradictory to the DSC findings but there are numerous issues under consideration. The most likely explanation is that the A6 – A11 cystathionine bridge diminishes the conformational stability of insulin but enhances the covalent stability of the cystine framework.

5.8 Conclusions

The A11 analogue was tested in both the IR-A and IR-B binding assays and was found to have the same affinity for both receptors as that of native human insulin. This is a promising initial result although not a surprising one since the only structural modification was the replacement of a sulfur atom with a methylene group. This outcome also confirmed that the correctly folded peptide

114 was synthesized and validated this method as a way of incorporating cystathionine in peptides. There was no discernible difference between the CD spectrum of A11 insulin and that of the native molecule, and the proportion of α- helicity was, therefore, calculated to be approximately the same. 1H NMR spectra were also obtained and, although the final structure has yet to be elucidated, close inspection of the data of both A11 and native insulin reveals significant similarity between the two. This suggests that both analogues are conformationally similar in three-dimensional space. This finding is consistent with the binding data results.

The plasma stability of A11 insulin was also tested in human serum, and it was found to be relatively stable after a period of 24 hours which is similar to the native hormone. This is an excellent result because even minor modifications of peptides can render them unstable, as slight conformational changes can expose amide linkages to proteases. The DSC analysis showed that key phase transitions occurred at slightly lower temperatures for A11 insulin than for the native, which indicates that the cystathionine bridge may destabilize the tertiary structure in a minor way. Finally, the thermal stability was tested and after incubation at 60 oC for 4 days, the native insulin degraded by more than 60% but the A11 analogue had only degraded by approximately 10%. This is an interesting result because the A7 – B7 disulfide bond is more prone to cleavage than A6 – A11.24 It is possible that the A6 cysteine which is vicinal to the A7 residue plays a role in the disulfide exchange during formation of the covalent oligomers. These preliminary results are encouraging and, although further evaluation is required, there is now strong indication that the A11 insulin cystathionine analogue is more thermally stable, thus enhancing its potential therapeutic usage.

115

5.9 References

(1) Hossain, M. A., Haugaard-Kedström, L. M., Rosengren, K. J., Bathgate, R. A. D., Wade, J. D. "Chemically Synthesized Dicarba H2 Relaxin Analogues Retain Strong RXFP1 Receptor Activity but Show an Unexpected Loss of in Vitro Serum Stability." Org Biomol Chem 2015, 13 (44), 10895–10903.

(2) Belfiore, A., Frasca, F., Pandini, G., Sciacca, L., Vigneri, R. "Insulin Receptor Isoforms and Insulin Receptor/Insulin-Like Growth Factor Receptor Hybrids in Physiology and Disease." Endocr. Rev. 2009, 30 (6), 586–623.

(3) De Meyts, P. "Insulin and Its Receptor: Structure, Function and Evolution." BioEssays 2004, 26 (12), 1351–1362.

(4) Denley, A., Bonython, E. R., Booker, G. W., Cosgrove, L. J., Forbes, B. E., Ward, C. W., Wallace, J. C. "Structural Determinants for High-Affinity Binding of Insulin-Like Growth Factor II to Insulin Receptor (IR)-A, the Exon 11 Minus Isoform of the IR." Mol. Endocrinol. 2004, 18 (10), 2502– 2512.

(5) Scholtz, J. M., Qian, H., York, E. J., Stewart, J. M., Baldwin, R. L. "Parameters of Helix-Coil Transition Theory for Alanine-Based Peptides of Varying Chain Lengths in Water." Biopolymers 1991, 31, 1463–1470.

(6) Wlodawer, A., Savage, H., Dodson, G. "Structure of Insulin: Results of Joint Neutron and X-Ray Refinement." Acta Crystallogr. Sect. B 1989, 45 (1), 99–107.

(7) Zhang, S., Hughes, R. A., Bathgate, R. A. D., Shabanpoor, F., Hossain, M. A., Lin, F., van Lierop, B., Robinson, A. J., Wade, J. D. "Role of the Intra-A-Chain Disulfide Bond of Insulin-like Peptide 3 in Binding and Activation of Its Receptor, RXFP2." Peptides 2010, 31 (9), 1730–1736.

116

(8) Hossain, M. A., Rosengren, K. J., Zhang, S., Bathgate, R. A. D., Tregear, G. W., van Lierop, B. J., Robinson, A. J., Wade, J. D. "Solid Phase Synthesis and Structural Analysis of Novel A-Chain Dicarba Analogs of Human Relaxin-3 (INSL7) That Exhibit Full Biological Activity." Org. Biomol. Chem. 2009, 7 (8), 1547.

(9) Williams, G. M., Lee, K., Li, X., Cooper, G. J. S., Brimble, M. A. "Replacement of the CysA7-CysB7 Disulfide Bond with a 1,2,3-Triazole Linker Causes Unfolding in Insulin Glargine." Org. Biomol. Chem. 2015, 13 (13), 4059–4063.

(10) Fredericq, E. "The Association of Insulin Molecular Units in Aqueous Solutions." Arch. Biochem. Biophys. 1956, 65, 218–228.

(11) Borowicz, P., Bocian, W., Sitkowski, J., Bednarek, E., Mikiewicz-Syguła, D., Kurzynoga, D., Stadnik, D., Surmacz-Chwedoruk, W., Koźmiński, W., Kozerski, L. "Biosynthetic Engineered B28K–B29P Human Insulin Monomer Structure in Water and in Water/acetonitrile Solutions." J. Biomol. NMR 2013, 55 (3), 303–309.

(12) Křížková, K., Veverka, V., Maletínská, L., Hexnerová, R., Brzozowski, A. M., Jiráček, J., Žáková, L. "Structural and Functional Study of the GlnB22- Insulin Mutant Responsible for Maturity-Onset Diabetes of the Young." PLoS ONE 2014, 9 (11), e112883.

(13) Aue, W. P., Bartholdi, E., Ernst, R. R. "Two-Dimensional Spectroscopy. Application to Nuclear Magnetic Resonance." J. Chem. Phys. 1976, 64 (5), 2229.

(14) Spink, C. H. "Differential Scanning Calorimetry." In Methods in Cell Biology, Elsevier, 2008, Vol. 84, pp 115–141.

(15) Huus, K., Havelund, S., Olsen, H. B., van de Weert, M., Frokjaer, S. "Chemical and Thermal Stability of Insulin: Effects of Zinc and Ligand Binding to the Insulin Zinc-Hexamer." Pharm. Res. 2006, 23 (11), 2611– 2620.

117

(16) Hua, Q. X., Ladbury, J. E., Weiss, M. A. "Dynamics of a Monomeric Insulin Analog: Testing the Molten-Globule Hypothesis." Biochemistry (Mosc.) 1993, 32 (6), 1433–1442.

(17) Duckworth, W. C., Bennett, R. G., Hamel, F. G. "Insulin Degradation: Progress and Potential." Endocr. Rev. 1998, 19 (5), 608–624.

(18) Duckworth, W. C. "Insulin Degradation: Mechanisms, Products, and Significance." Endocr. Rev. 1988, 9 (3), 319–345.

(19) Gambhir, K. K., Nerurkar, S. G., Das, P. D., Archer, J. A., Henry JR, W. L. "Insulin Binding and Degradation by Human Erythrocytes at Physiological Temperature." Endocrinology 1981, 109 (5), 1787–1789.

(20) Livesey, J. H., Hodgkinson, S. C., Roud, H. R., Donald, R. A. "Effect of Time, Temperature and Freezing on the Stability of Immunoreactive LH, FSH, TSH, Growth Hormone, Prolactin and Insulin in Plasma." Clin. Biochem. 1980, 13 (4), 151–155.

(21) Knappe, D., Henklein, P., Hoffmann, R., Hilpert, K. "Easy Strategy To Protect Antimicrobial Peptides from Fast Degradation in Serum." Antimicrob. Agents Chemother. 2010, 54 (9), 4003–4005.

(22) Phillips, N. B., Whittaker, J., Ismail-Beigi, F., Weiss, M. A. "Insulin Fibrillation and Protein Design: Topological Resistance of Single-Chain Analogs to Thermal Degradation with Application to a Pump Reservoir." J. Diabetes Sci. Technol. 2012, 6 (2), 277–288.

(23) Dantas de Araujo, A., Mobli, M., Castro, J., Harrington, A. M., Vetter, I., Dekan, Z., Muttenhaler, M., Wan, J., Lewis, R. J., King, G. F., Brierley, S. M., Alewood, P. F. "Selenoether Oxytocin Analogues Have Analgesic Properties in a Mouse Model of Chronic Abdominal Pain." Nat. Commun. 2014, 5, 3165.

118

(24) Brange, J., Havelund, S., Hougaard, P. Chemical Stability of Insulin 2. "Formation of Higher Molecular Weight Transformation Products During Storage of Pharmaceutical Preparations." Pharm. Res. 1992, 9 (6), 727– 734.

119

120

Chapter 6

Conclusions and Future Work

121

6.1 Introduction

Insulin-like peptides are challenging molecules to work with from a synthetic point of view due to their heterodimeric nature and three cystine moieties. Nevertheless improved methods have been developed during the course of this work that addresses some of the difficulties. Firstly, the synthetic yield for a key step in the regio-selective formation of one of the disulfide bonds has been increased. Furthermore the solubility of the intermediate species was enhanced which also contributed to the higher yields. Finally, new synthetic routes were established which enabled the convenient incorporation of non-reducible cystine isosteres into insulin-like peptides. These advances make the development of second-generation therapeutics with enhanced physico-chemical and pharmacological properties more feasible from this peptide family, especially for human insulin. A concise summary of both the newly developed synthetic methods and the substitution of cystine with stable mimetics follows.

6.2 Synthesis of Insulin via 2-Nitroveratryl Thiol Protection

Given the inherent difficulties in forming multiple disulfide bonds in cystine-rich peptides such as insulin, a new protecting group strategy was developed that resulted in enhanced yields (Chapter 2). This could allow peptide analogues with non-native moieties to be feasible as lead compounds because a chemical synthesis can then be employed. In order to avoid performing an often times low-yielding oxidative cleavage of two S-Acm protected cysteine residues with iodine, a photocleavable S-protecting group was used instead. This is advantageous since it can be cleaved under ambient conditions and hence minimizes the formation of side products. S-oNv was incorporated into peptides and its cysteine “partner” was functionalized with S-Pyr. This allows efficient thiolysis to occur after photolytic cleavage and reduces unwanted disulfide shuffling.

To incorporate S-oNv into peptides, the building block Fmoc-Cys(oNv)-OH was synthesized in good yield. A flash purification method was established, the

122 molecule was fully characterized and the coupling conditions were optimized to minimized epimerization. Oxytocin was the first model system used whereby S- tBu and S-oNv protection was incorporated respectively at positions 1 and 6. Cleavage of the former with TFMSA and concomitant functionalization to S-Pyr was efficient and the major product formed after photolysis was the native hormone (Chapter 2). A similar approach was then applied to α-conotoxin ImI which has two disulfide bonds. Both cystine moieties were formed efficiently with no evidence of disulfide scrambling.

This method was then applied to insulin. After assembly, purification and dimerization of both chains, B7 was S-Pyr functionalized followed by photolysis and thiolysis to form the A7 – B7 disulfide bond. A temporary iso-acyl bond was incorporated into the backbone and significantly improved solubility. The final product was fully chemically characterized and had native affinity for the insulin receptor. The yield for the formation of the final disulfide bond was 18% which is superior to previously reported methods using iodine.1 The total synthetic yield calculated form B-chain was 7% (Chapter 2). Although it doesn’t match recombinant yields, this result is nevertheless promising and there is much scope for further optimization.

In the future, a similar approach can be applied to relaxin-2 given that this peptide is in Phase IIIb clinical trials. Furthermore it contains sensitive residues – two methionine and two tryptophan residues – so would, therefore, be a good candidate to compare iodine oxidation yields with those derived from the photochemical approach. This method can also be applied to other insulin-like peptides and disulfide-rich biomolecules in order to demonstrate the versatility and utility of this methodology. There is also the possibility of simplifying disulfide bond formation by investigating the use of an S-oNv/S-oNv pairing instead of an S-oNv/S-Pyr combination (Figure 6.1(i)). This would eliminate one synthetic step: the treatment of the S-tBu protected peptide with 10% TFMSA/TFA and DPDS. Optimization is required and could be accomplished by improving the rate of photolysis, such that both S-oNv moieties are cleaved “simultaneously” to ensure that the desired disulfide bond is formed and

123 scrambled isomers are kept to a minimum. Furthermore, this reaction should be performed in the presence of DPDS with optimal stoichiometric ratios. There will be a trade-off between using enough reagent to afford fast conversion to the disulfide without generating significant pyridinesulfenyl adducts. The use of photochemistry can be extended to other protecting groups such as S-7- methoxycoumaryls which are semi-orthogonal to 2-nitroveratryl and can be selectively cleaved at longer wavelengths such as 430 nm.2 This could allow peptides with three or more disulfide bonds to be efficiently synthesized regioselectively. A proposed synthetic strategy is highlighted in Figure 6.1(ii), with ω-Conotoxin MVIIC as the model system. Finally, S-oNv can be used to protect N-terminal cysteine fragments for use in one-pot native chemical ligation reactions.3 Employing photocleavable protecting groups for the purpose of forming cystine in peptides has proven to be a robust and versatile technique, and there is much scope to further improve upon this approach.

124

6.3 Incorporation of Stable Cystine Isosteres in Insulin-like Peptides

One way to improve the physico-chemical and pharmacological properties of cystine-rich peptides is to replace the disulfide bonds with more stable mimetics.4 More specifically, this approach can be used to synthesize improved insulin analogues with enhanced thermal stability. But this is particularly challenging for insulin-like peptides because, although techniques for incorporating cystine isosteres in single chain peptides have been well established, replacement of the intermolecular bridges has proved more difficult. Therefore, a proof-of-concept project was pursued whereby lactam bridges were incorporated into INSL3. The rationale behind these choices was firstly because INSL3 is more water-soluble and thus easier to purify, and, secondly, the building blocks required to form the lactam bridges are commercially available (Chapter 3). This was followed by developing new chemical methods for incorporating cystathionine bridges into peptides, which are more isosterically similar to cystine, and applied to human insulin given its therapeutic relevance (Chapter 4). The yield of each synthetic step was particularly important for this work since any lead compounds developed should be scalable, otherwise it will not be viable as a therapeutic. After synthesis the peptide was then be characterized by binding affinity to the insulin receptor and its serum and thermal stability can be assessed (Chapter 5).

A novel approach was adopted to incorporate Dpr/Asp intermolecular lactam bridges into INSL3 (ΔA25/A26) which involved using protecting groups orthogonal to Fmoc/tBu protocols and assembling both chains on the same solid support. The first peptide synthesized was the A24 – B22 mono-lactam analogue, whereby the B-chain was assembled followed by the A-chain via amide attachment at B22 (Chapter 3). Each disulfide bond was formed sequentially and the final product was purified and characterized, and an overall yield of 0.2% was obtained. A similar approach was used for a bis-lactam analogue with modifications at both A11 – B10 and A24 – B22, but in this instance the macrocyclization occurred on the solid support. The peptide was oxidized, purified and characterized but an extremely low yield of less than

125

0.01% was obtained, which is indicative of the complexity of the synthesis. This is likely due to an unfavorable conformation on the solid support, steric hindrance, aggregation and cross-linking.

Although this synthetic approach is viable at the discovery stage of drug development, it is not scalable for downstream applications so an alternative method should be pursued that involves ligation of two or more fragments. Nevertheless, receptor binding data was obtained which showed that the mono- lactam analogue had virtually the same affinity for the RXFP2 receptor as the INSL3 (ΔA25/A26) analogue. The bis-lactam analogue had a slightly reduced affinity which is probably due to the structural and electronic differences between the amide and disulfide bonds.

Given the synthetic difficulties and structural differences in the examples above, it was thought that cystathionine should be an improvement on lactam bridges. Because alternative chemical protocols are required to ensure higher yields, new building blocks were employed that are easy to synthesize. The A11 insulin analogue (A6 – A11 cystathionine bridge with γ-methylene at A11) was deemed to be a suitable first target to produce because incorporating disulfide mimics in single chain peptides is simpler. The required orthogonally protected building blocks Fmoc-γ-Br-hAla-OH and Mmt-Cys-OAll were prepared in decent yields of 84% and 80%, respectively (Chapter 4). The halogenated residue was incorporated at position A11 on the solid support, followed by alkylation via an

SN2 reaction to form cystathionine with the introduction of the cysteine derivative. After chain elongation up to residue A7 and deprotection, the macrocycle was formed via a lactamization. Assembly of the A-chain was completed followed by isolation and purification. Each chemical step resulted in one major product which indicates that this approach is effective and further optimization should be possible. In this instance, the final yield of the A-chain analogue was 12%. The B-chain was derivatized with S-Pyr and purified followed by chain combination and an iodine oxidation to generate A11 insulin, with a folding yield of 11%.

126

A11 insulin was found to have native binding affinity, which further validates this method as a way of incorporating cystathionine in peptides. The secondary structure was almost identical to that of the native peptide as determined by CD spectroscopy. 1H NMR spectra were also obtained and there was a great deal of similarity between the two indicating that both proteins should have almost identical tertiary structures. The human serum stability was evaluated and it was found that the cystathionine bridge did not significantly alter the half-life in vitro. This is an important result because an unstable analogue would be unviable as a therapeutic. DSC analysis indicated that the A11 analogue was slightly more susceptible to conformational changes during heating than native insulin. Most importantly though, the thermal stability was evaluated at 60 oC and it was found that native insulin degraded significantly after 4 days whereas the A11 analogue was relatively stable. This is a promising outcome and could pave the way for heat-stable insulins that are less prone to degradation when stored at elevated temperatures, a problem that occurs frequently in developing countries with tropical climates.

Future work in this area will involve further characterization of the A11 insulin analogue, firstly by solving the solution-state NMR structure and then testing it in vivo to determine whether it has the same blood-glucose lowering effect as native insulin. The binding affinity for the IGF-II receptor should also be measured as well as the redox stability. The thermal stability of the zinc-bound hexameric form should also be evaluated in order to more closely mimic insulin formulations. In addition, A20 – B19 and A7 – B7 cystathionine analogues of insulin should be synthesized and characterized in a similar fashion to determine which one has the optimal physico-chemical and pharmacological properties. Assembly of these analogues will be challenging but many valuable insights were gained from incorporating the intermolecular lactam bridges in INSL3 which can aid in the design of better synthetic routes.

The A20 – B19 cystathionine analogue can be assembled in a similar fashion to the INSL3 analogues but using Fmoc-γ-Br-hAla-OH and Mmt-Cys-OAll to form the C-terminal intermolecular bridge (Figure 6.2). The allyl group should then be

127 cleaved followed by incorporation of H-Asn(Trt)-OtBu.HCl at the C-terminus, which can be synthesized with commercially available starting materials. In order to simplify the synthesis, the B-chain should be assembled up to residue B7 with N-terminal Boc-protected thiazolidine which acts as a surrogate cysteine residue.3 The A-chain can then be assembled up to residue A7 with an N-terminal cysteine followed by TFA cleavage and purification. A1 – A6 and B1 – B6 fragments should then be prepared with C-terminal hydrazides followed by oxidation to the acyl azide and subsequent formation of a thioester.5 The N- terminal A-chain fragment can then be attached via native chemical ligation (NCL), followed by ring opening of thiazolidine with methoxyamine3 and attachment of the B-chain via NCL. A6 and A11 cysteine residues could be S- Acm protected such that a one-pot two-step iodine oxidation6 can be undertaken to form the desired analogue.

The A7 – B7 analogue is synthetically more challenging since it is difficult to incorporate an intermolecular bridge near the N-terminus. Nevertheless this has previously been achieved with INSL3, whereby two fragments were prepared and ligated together.7 The A-chain could be assembled up to A8, followed by incorporation of Fmoc-γ-Br-hAla-OH and Mmt-Cys-OAll to form the thioether bridge. The allyl group can again be cleaved followed by incorporation of H-Gly- NHNH-Boc.HCl which can be synthesized with commercially available reagents and incorporates a hydrazide moiety at position B8. A-chain assembly can continue followed by capping with a hydrazine-labile dimedone-based PEG- ylated protecting group which will both block the α-amino group and enhance the solubility of the peptide construct in aqueous media. The B-chain can then be extended and capped using the same protection strategy and S-Acm at positions A6 and A11, followed by TFA cleavage and purification. Fragment B9 – B30 can then be prepared with dimedone-based protection on the lysine side- chain at B29 and S-pyridinesulfenyl functionalization at B19. After chain combination at pH 8 and purification, the larger fragment can then be oxidized from the hydrazide to the azide and then used to directly acylate the B9 – B30 fragment. This can be followed by oxidative cleavage and deprotection with hydrazine to form the desired analogue (see Figure 6.3).

128

129

130

These approaches will generate B19 and A7 “carba” analogues of insulin given that the cystathionine bridge is asymmetrical. These methods can be extended to form A6, A20 and B7 γ-carba insulins by simply using different building blocks: Fmoc-β-Br-Ala-OH and Mmt-hCys-OAll. Furthermore there is the potential to generate bis-cystathionine and tris-cystathionine analogues since replacement of more disulfide bonds in insulin could further enhance the thermal (and possibly metabolic) stability, which could lead to the development of an enhanced therapeutic.

6.4 Summary

New chemical methods for the synthesis of insulin-like peptides have been described. Higher yields were achieved for the formation of disulfide bonds in insulin by employing 2-nitroveratryl as a photocleavable thiol protecting group. Non-reducible cystine isosteres were also incorporated into members of this peptide family. A cystathionine-containing insulin analogue with native receptor binding affinity was synthesized in excellent yield and had enhanced thermal stability. This analogue has therapeutic potential, especially in the developing world where refrigeration is not ubiquitous. Further research directions based on these findings are proposed.

131

6.5 References

(1) Akhter Hossain, M., Bathgate, R. A. D., Kong, C. K., Shabanpoor, F., Zhang, S., Haugaard-Jönsson, 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 (11), 1816–1822.

(2) Kotzur, N., Briand, B., Beyermann, M., Hagen, V. "Wavelength-Selective Photoactivatable Protecting Groups for Thiols." J. Am. Chem. Soc. 2009, 131 (46), 16927–16931.

(3) Bang, D., Kent, S. B. H. "A One-Pot Total Synthesis of Crambin." Angew. Chem. Int. Ed. 2004, 43 (19), 2534–2538.

(4) Dekan, Z., Vetter, I., Daly, N. L., Craik, D. J., Lewis, R. J., Alewood, P. F. "α-Conotoxin ImI Incorporating Stable Cystathionine Bridges Maintains Full Potency and Identical Three-Dimensional Structure." J. Am. Chem. Soc. 2011, 133, 15866–15869.

(5) Fang, G.-M., Li, Y.-M., Shen, F., Huang, Y.-C., Li, J.-B., Lin, Y., Cui, H.-K., Liu, L. "Protein Chemical Synthesis by Ligation of Peptide Hydrazides." Angew. Chem. 2011, 123 (33), 7787–7791.

(6) Townsend, A., Livett, B. G., Bingham, J.-P., Truong, H.-T., Karas, J. A., O’Donnell, P., Williamson, N. A., Purcell, A. W., Scanlon, D. "Mass Spectral Identification of Vc1.1 and Differential Distribution of Conopeptides in the Venom Duct of Conus Victoriae. Effect of Post- Translational Modifications and Disulfide Isomerisation on Bioactivity." Int. J. Pept. Res. Ther. 2009, 15 (3), 195–203.

(7) Büllesbach, E. E., Schwabe, C. "Replacement of Disulfides by Amide Bonds in the Relaxin-like Factor (RLF/INSL3) Reveals a Role for the A11– B10 Link in Transmembrane Signaling." Biochemistry (Mosc.) 2012, 51 (20), 4198–4205.

132

133

Appendix 1

Supporting Information for

Chapter 2

134

Table of Contents

1. Solid phase peptide synthesis and reagents ...... 136

2. Reversed Phase – HPLC ...... 136

3. Mass Spectrometry ...... 137

4. NMR Spectroscopy ...... 137

5. Polarimetry ...... 137

6. Photochemistry ...... 137

7. Fluorescence measurement of α7 nAChR activity...... 137

8. Insulin receptor binding assay ...... 138

9. Synthesis of Fmoc-Cys(oNv)-OH ...... 138

10. Fmoc-Cys(oNv)-OH racemization study ...... 141

11. Synthesis of Oxytocin ...... 142

12. Synthesis of α-Conotoxin ImI (1) ...... 144

13. Synthesis of α-Conotoxin ImI (2) ...... 145

14. Synthesis of Insulin (1) ...... 147

15. Synthesis of Insulin (2) ...... 148

135

1. Solid phase peptide synthesis and reagents

All peptides were assembled on a CEM Liberty microwave peptide synthesizer. Fmoc-Cys(oNv)-OH and Boc-Ser(Fmoc-Thr(tBu))-OH were coupled manually, using HOBt/DIC coupling conditions. Standard Fmoc amino acids, Fmoc-Asp- OtBu, Fmoc-Cys(tBu)-OH, HATU, HOBt and Boc anhydride were obtained from GL Biochem (China). Fmoc-Cys-OH was obtained from Iris Biotech (Germany). Boc-Ser(Fmoc-Thr(tBu))-OH, Fmoc-Cys(Mmt)-OH, diethyl ether, acetonitrile, acetic acid, sodium hydroxide, methanol, DIEA, DMF piperidine and guanidinium hydrochloride were sourced from Merck Millipore (Australia). NMP was obtained from Applied Biosystems (USA). TFA and dichloromethane were purchased from Auspep Pty Ltd (Australia). 4,5-dimethoxy-2-nitrobenzyl bromide, DIC, TIPS, DODT, TFMSA and thioanisole were obtained from Sigma- Aldrich (Australia). Tentagel MB RAM (sub = 0.26 mmol/g) and Tentagel R-PHB Fmoc-Thr(tBu) (sub = 0.18 mmol/g) resins were sourced from Rapp Polymere (Germany).

2. Reversed Phase – HPLC

All RP-HPLC was done on a Waters 600 HPLC controller with a 600 pump and a 996 photodiode array detector. Mobile phase was as follows: buffer A = 0.1% TFA in water; buffer B = 0.1% TFA in acetonitrile. Analysis was performed on a Phenomenex Kinetex 5µ XB-C18, 100 Å, 50 x 4.6 mm column. A typical gradient used was 10-70% buffer B over 30 minutes with a 1.5 mL/min flow rate and detection at 210 nm. Peptide purification was carried out on a Phenomenex Kinetex 5µ XB-C18, 100 Å, 150 x 21.2 mm AXIA packed column. A typical gradient used was 20-60% buffer B over 40 minutes with a 10 mL/min flow rate and detection at 230 nm. All isolated products were lyophilized in a Christ freeze dryer.

136

3. Mass Spectrometry

MALDI-MS experiments were performed on a Bruker Daltonics Autoflex II TOF/TOF using sinapinic acid as the matrix. ESI-MS experiments were performed on an Agilent 6220 Accurate-Mass TOF LC/MS.

4. NMR Spectroscopy

1H and 13C NMR experiments were performed on a 400 MHz Agilent MR 400 nuclear magnetic resonance spectrometer in deuterated water.

5. Polarimetry

Polarimetry measurements were obtained on a Jasco DIP-1000 Digital Polarimeter with chloroform as the solvent.

6. Photochemistry

All photolysis reactions were performed in the 350 nm range with a RPR-100 Rayonet Photochemical Chamber Reactor.

7. Fluorescence measurement of α7 nAChR activity

Fluorescence measurement of α7 nAChR activity was performed using the FLIPRTETRA fluorescence plate reader (Molecular Devices, Sunnyvale, CA). SH- SY5Y human neuroblastoma cells (European Cell Culture Collection) were cultured in RPMI medium (Invitrogen, Australia) supplemented with 15% foetal bovine serum and L-glutamine (1 mM). Cells were plated at a density of 30,000 cells/well on 384-well black-walled imaging plates (Corning) 48 h prior to FLIPR assays and loaded with Calcium 4 no-wash dye (Molecular Devices) diluted in

137 physiological salt solution (PSS; composition in mM: NaCl 140, glucose 11.5,

KCl 5.9, MgCl2 1.4, NaH2PO4 1.2, NaHCO3 5, CaCl2 1.8, HEPES 10) for 30 min at 37°C. Fluorescence responses (excitation 470–495 nm; emission 515–575 nm) to stimulation with the α7 nAChR agonist choline (30 µM) and PNU120596 (10 µM, Sigma Aldrich) were assessed after 5 min pre-treatment with ImI using the FLIPRTetra+ fluorescence plate reader (Molecular Devices). Synthetic ImI, a kind gift of Prof Paul Alewood (Institute for Molecular Bioscience, The University of Queensland), was used as a positive control. Raw fluorescence readings were converted to response over baseline using Screenworks 3.1.1.4 (Molecular Devices) and were expressed relative to the maximum increase in fluorescence of control responses. A minimum of 3 wells/concentration were used to determine the IC50 in 5 independent experiments. Performed by Dr Irina Vetter.

8. Insulin receptor binding assay

Receptor binding was measured by generating IGF-1R-negative cells overexpressing the IR-B (R-IR-B). Cells were serum-starved for 4h before lysis. Lysates were captured in a 96 well plate previously coated with anti-IR antibody 83-7. Approximately 500,000 fluorescent counts of europium labelled insulin were added to each well along with increasing concentrations of unlabelled competitor and incubated for 16 h at 4oC. After washing time-resolved fluorescence was measured using 340 nm excitation and 612 nm emission filters with a BMG Lab technologies Polarstar fluorometer (Mornington, Australia). Assays were performed in triplicate three times. This assay was performed by Assoc. Prof. Briony Forbes.

9. Synthesis of Fmoc-Cys(oNv)-OH

Fmoc-Cys(oNv)-OH was synthesized as outlined in Figure A1.1. 4,5-Dimethoxy- 2-nitrobenzyl bromide (0.5 mmol, 138 mg) was dissolved in DCM (5 mL, 100

138 mM). DIEA (2.5 mmol, 435 µL) was added drop-wise, followed by the slow addition of Fmoc-Cys-OH (0.5 mmol, 172 mg). The reaction vessel was covered in foil and stirred for 18 hours. The solution was reduced in vacuo and the oily residue was re-dissolved in DMF (2.5 mL), then added drop-wise to a vigorously stirred aqueous solution of cold HCl (0.5 N, 50 mL). The yellow precipitate was filtered then washed with water (3 x 20 mL) and dried at the pump. The product was purified via flash chromatography (0 – 10% MeOH in DCM). 186 mg of yellow powder was recovered, indicating a yield of 69%. All data presented in Figure A1.2. Melting point = 202 – 206 oC. Specific rotation: [α] = -78.7o. 1H NMR (400 MHz, DMSO-d6): δ = 7.87 (d, 2H, J = 7.5 Hz), 7.71 (t, 2H, J = 8.1 Hz), 7.63 (s, 1H), 7.39 (t, 2H, J = 7.2 Hz), 7.30 (t, 2H, J = 7.4 Hz), 4.29-4.18 (m, 2H), 4.15-4.02 (m, 2H), 3.85 (s, 3H), 3.81 (s, 3H), 2.85 (dd, J = 13.9 Hz, 4.6 Hz, 1H), 2.69 (dd, J = 13.9 Hz, 9.7 Hz, 1H), 2.48 (t, J = 3.7 Hz, 4H), 2.05. 13C NMR (400 MHz, DMSO-d6): δ = 172.69, 156.46, 153.01, 147.88, 144.24, 144.21, 141.15, 140.24, 129.38, 128.09, 127.52, 125.74, 125.70, 120.56, 114.66, 109.31, 66.18, 56.61, 56.44, 54.31, 47.04. Mono-isotopic molecular mass of the [M + H+]+ species via ESI-MS analysis = 538.13; theory = 538.14.

139

140

10. Fmoc-Cys(oNv)-OH Racemization study

The GCF model peptide was prepared manually using standard SPPS techniques Three activation protocols were evaluated for Fmoc-Cys(oNv)-OH: (A) HATU/DIEA at ambient temperature, (B) HATU/DIEA at 60oC and (C) HOBt/DIC at ambient temperature. Activation times were 5 minutes. The peptides were cleaved from the resin with a cocktail of TIPS/H2O/TFA (1%/2%/97%). The LL and DL diastereoisomers were detected via analytical RP-HPLC using a gradient of 10-50% buffer B over 40 minutes (Figure A1.3). Levels of epimerization were calculated as per the following formula: 100x[Area(D)/(Area(L)+Area(D))]. The impurity peak was ignored. The results were as follows: (A) HATU/DIEA at 25oC = 4% DL; (B) HATU/DIEA at 60oC = 22% DL; (C) HOBt/DIC < 0.5% DL.

141

11. Synthesis of Oxytocin

The peptide (0.1 mmol) was assembled as outlined in Figure A1.4, using a Liberty microwave peptide synthesizer on Tentagel AM RAM resin. The N- terminus was Boc-protected via treatment with 5 equivalents of boc anhydride (0.5 mmol, 109.1mg) and DIEA (0.5 mmol, 85 µL) in DMF (I), followed by treatments with a solution of 1% TFA/DCM (10 x 5 mL for 2 minutes each). After neutralization of the resin with 5% DIEA/DMF, the resin bound peptide was treated with 3 equivalents of 4,5-dimethoxy-2-nitrobenxyl bromide (0.3 mmol,

82.8 mg) and then the peptide was cleaved with a cocktail of TIPS/H2O/TFA (1%/2%/2%/95%) and isolated as per standard SPPS protocols (II). 102.3 mg of crude pale yellow powder was recovered. 20.0 mg of this material was treated with DPDS (5 equivalents, 17.5 mg) and thioanisole (10 equivalents, 9.3 µL) in a 1 mL solution of 10% TFMSA/TFA at 0 oC for 30 minutes, followed by ether precipitation, centrifugation, RP-HPLC purification and lyophilization. 11.0 mg was recovered, indicating a yield of 53% (III). 3.0 mg of the intermediate was then dissolved in 6 mL of 1% acetic acid in 50% aqueous acetonitrile and irradiated at a wavelength of 350 nm for 30 minutes. The solution turned slightly brown and RP-HPLC and ESI-MS analysis confirmed that the reaction was complete and that native oxytocin had been synthesized (IV). The solution was injected directly on to the RP-HPLC column and isolated as per the normal protocols and 1.5 mg of material at >97% purity was recovered, indicating a yield of 65%. Total yield for the formation of the disulfide bond was 34%. ESI- MS mono-isotopic [M+H+]+ ion: experimental = 1007.44; theoretical = 1007.44. See Figure A1.5 for analytical data.

142

143

12. Synthesis of α-Conotoxin ImI (1)

α-Conotoxin ImI was assembled as outlined in Figure A1.6, on a Liberty peptide synthesizer using 5 eq. of HATU/DIPEA on a 100 µmol scale using Tentagel AM RAM (Rapp Polymere). Cys6 was incorporated with mono-methoxytrityl protection and the N-terminus was Boc protected with 10 eq. of Boc2O and DIEA (V*). The Cys6(Mmt) residue was deprotected with 1% TFA/DCM and reprotected with 3 eq. of 4,5-dimethoxy-2-nitrobenzyl bromide (0.3 mmol, 161.4 mg) and DIEA (51 µL) in DMF. Trial cleavages of the resin before and after the

2-nitroveratryl reaction were performed with TIPS/H2O/DODT/TFA (1%/2%/2%/95%) for 2 hours and ether precipitated (VI*). A number of 2- nitroveratryl adducts were observed as detected by RP-HPLC and MALDI-MS (Figure A1.7). This is most likely due to partial cleavage of S-Trt protected cysteines and subsequent reprotection of the free thiolates. S-Mmt cleavage was repeated using 0.5%TFA in DCM however a similar result occurred, therefore the synthesis was aborted.

144

13. Synthesis of α-Conotoxin ImI (2)

The peptide (0.05 mmol) was assembled as outlined in Figure A1.8 using a Liberty microwave peptide synthesizer with HATU/DIEA activation on Tentagel AM RAM resin up until Cys3, followed by manual coupling of Fmoc-Cys(oNv)- OH (0.1 mmol, 53.8 mg) via HOBt/DIC activation, and subsequent glycine addition. The peptide was then cleaved with a cocktail of TIPS/H2O/DODT/TFA (1%/2%/2%/95%) and isolated as per standard SPPS protocols. 38.0 mg of crude pale yellow powder was recovered, and purified via RP-HPLC (V). 18.7 mg of purified material was dissolved in 19 mL of a 10% aqueous acetonitrile and treated with DPDS (1 equivalent, 2.6 mg) followed by stirring for 30 minutes. The oxidized product was then repurified and 13.2 mg of semi-pure material was isolated (VI). The peptide was then treated with DPDS (5 equivalents, 9.1 mg) in a 1 mL solution of 10% TFMSA/TFA at 0oC for 30 minutes, followed by ether precipitation, centrifugation, RP-HPLC purification and lyophilization. 8.6 mg was recovered, indicating a yield of 63% (VII). 2.7 mg of the intermediate was then dissolved in 5.4 mL of 1% acetic acid in 50% aqueous acetonitrile and irradiated at a wavelength of 350 nm for 30 minutes. The solution turned slightly brown and RP-HPLC and ESI-MS analysis confirmed that the reaction was complete and that native α-conotoxin-ImI had been synthesized (VIII). The solution was injected directly on to the RP-HPLC column and isolated as per the normal protocols and 1.2 mg of material >98% pure was recovered, indicating a yield of 54%. Total yield for the formation of the disulfide bond was 34%. MALDI-MS average [M+H+]+ ion: experimental = 1352.1; theoretical = 1352.6. See Figure A1.9 for analytical data.

145

146

14. Synthesis of human Insulin (1)

A-chain and B-chain were assembled on a Liberty peptide synthesizer using 5 eq. of HATU/DIPEA on a 100 µmol scale using the relevant PEG-PS resin from Applied Biosystems. Fmoc-Cys(oNv)-OH was incorporated manually using 2 equivalents. Peptides were cleaved with TIPS/H2O/DODT/TFA (1%/2%/2%/95%) for 2 hours and ether precipitated. 110 mg of crude material was dissolved in 0.1% DIPEA, 20% MeCN/H2O at 1 mg/mL and 2 equivalents of DPDS were dissolved in a small volume of acetonitrile and added dropwise. After 30 minutes the reaction mixture was filtered, pumped onto a C18 column and purified using standard TFA buffers. The remaining 110 mg of crude linear was cyclized and purified at pH 2. The semipure cyclic A chain (tBu7, oNv20) material was combined to give 38 mg, data below. 20.4 mg of peptide was then treated with 9.1 mg of DPDS (5eq.) and 15 µL of thioanisole (10 eq.) in 1.5 mL of 10%TFMSA/TFA for 15 minutes at 0oC. The peptide was ether precipitated and centrifuged. 5.7 mg was recovered from the main fraction, data below: A chain (5.7 mg in 1200 µL of 6M Gn.HCL, pH 8.5) and B chain (7.4 mg in 400 µL of 6M Gn.HCl, pH 6.0) were combined and reacted for 15 minutes. 4.8 mg of the dimer was recovered, which was then treated with 0.9 mg of DPDS (5 eq.) in 1 mL of 10% TFMSA/TFA for 15 minutes at 0oC. The peptide was ether

147 precipitated, centrifuged then analyzed and some purified material (approximately 0.5 mg) was recovered post purification (XIII*). The peptide was dissolved in 5 mL of 20% MeCN/H2O with 1% AcOH, and irradiated with hν at a wavelength of 350 nm for 45 minutes (Figure A1.10). The reaction was followed by RP-HPLC (Figure A1.11) and characterized by MALDI-MS. The same molecular weight was observed for the later eluting peaks (unpublished data), which suggests that scrambling of the disulfides occurs. The material was purified and isolated (XV*). Final mass was <50 µg.

15. Synthesis of human Insulin (2)

The A-chain peptide (0.125 mmol) was assembled as outlined in Figure A1.12, using a Liberty microwave peptide synthesizer with HATU/DIEA activation on Tentagel AM RAM resin up until Ile10, followed by manual coupling of Boc- Ser(Fmoc-Thr(tBu))-OH (0.2 mmol, 116.9 mg) and Fmoc-Cys(oNv)-OH (0.2 mmol, 107.6 mg) via HOBt/DIC activation, and subsequent addition of the N- terminal amino acids. The peptide was then cleaved with a cocktail of

TIPS/H2O/DODT/TFA (1%/2%/2%/95%) and isolated as per standard SPPS protocols. 528.1 mg of crude pale yellow powder was recovered (IX). 255.0 mg of crude material was dissolved in 255 mL of 10% aqueous acetonitrile and treated with DPDS (1 equivalent, 21.3 mg) followed by stirring for 30 minutes. The oxidized product was then purified and 57.8 mg of semi-pure material was isolated (X). 44.1 mg of the peptide was then treated with DPDS (5 equivalents, 18.4 mg) and thioanisole (10 equivalents, 21.2 µL) in a 2 mL solution of 10% TFMSA/TFA at 0 oC for 30 minutes, followed by ether precipitation, centrifugation, RP-HPLC purification and lyophilization. 22.1 mg was recovered, indicating a yield of 63% (XI). The B-chain peptide (0.125 mmol) was assembled using a Liberty microwave peptide synthesizer with HATU/DIEA activation on Tentagel R-PHB Fmoc-Thr(tBu) resin, then cleaved with a cocktail of TIPS/H2O/DODT/TFA (1%/2%/2%/95%) and isolated as per standard SPPS protocols (XI*). The yield post purification was 109.2 mg. 22.1 mg of A-chain

148 and 28.5 mg of B-chain was dissolved in 8mL of 6M solution of Gn.HCl and 5% acetonitrile, and stirred for 45 minutes, followed by purification (XII). 22.8mg of hetero-dimer recovered was then treated with DPDS (5 equivalents, 4.1 mg) and thioanisole (10 equivalents, 4.4 µL) in a 3 mL solution of 10% TFMSA/TFA at 0oC for 30 minutes, followed by ether precipitation, centrifugation, RP-HPLC purification and lyophilization (XIII). 11.4 mg was recovered, indicating a yield of 50%. 5.5 mg of the intermediate was then dissolved in 22 mL of 6M Gn.HCl at pH 5.5 and irradiated at a wavelength of 350 nm for 45 minutes. The solution turned slightly brown and RP-HPLC and ESI-MS analysis confirmed that the reaction was complete and that the insulin iso-acyl peptide had been formed (XIV). The pH was adjusted to 7.5 with a 0.1 M solution of NaOH. A shift in retention time was observed in the RP-HPLC chromatogram indicated that the

149

ON acyl shift between ThrA8 and SerA9 had occurred, and that native insulin had been synthesized (XV). The solution was injected directly on to the RP- HPLC column and isolated as per the normal protocols and 1.9 mg of material >96% pure was recovered, indicating a yield of 36%. Total yield for the formation of the disulfide bond was 18%. MALDI-MS average [M+H+]+ ion: experimental = 5807.6; theoretical = 5808.7. See Figures A1.13 and A1.14 for analytical data.

150

151

152

Appendix 2

Supporting Information for

Chapters 5 and 6

153

Table of Contents

1. Solid phase peptide synthesis and reagents ...... 155

2. Reversed Phase – HPLC ...... 155

3. Mass Spectrometry ...... 156

4. Amino Acid Analysis ...... 156

5. NMR Spectroscopy ...... 156

6. Synthesis of Fmoc-γ-Br-hAla-OH ...... 156

7. Synthesis of Mmt-Cys-OAll ...... 158

8. Synthesis of A11 Insulin ...... 159

9. Conversion of Recombinant Human Insulin to the TFA Salt...... 160

10. Insulin Receptor Binding Assays ...... 161

11. Circular Dichroism Spectroscopy ...... 161

12. Differential Scanning Calorimetry ...... 161

13. Human Serum Stability Assay ...... 161

14. Thermal Stability Assay ...... 162

154

1. Solid phase peptide synthesis and reagents

All amino acids were incorporated on to the peptide chain with a CEM Liberty microwave peptide synthesizer, except for Fmoc-γ-Br-hAla-OH and Mmt-Cys- OAll which were coupled manually. Tentagel R-PHB Fmoc-Thr(tBu) (sub = 0.18 mmol/g) was sourced from Rapp Polymere (Germany). Polystyrene rink amide resin (sub = 0.35 mmol/g), standard Fmoc amino acids, Fmoc-Asp-OtBu, Fmoc- Cys(Acm)-OH, Fmoc-OSu, HATU and HOBt were obtained from GL Biochem (China). Fmoc-Cys(Dpm)-OH was obtained from Iris Biotech (Germany). Silica gel 60, diethyl ether, acetonitrile, acetic acid, 33% aqueous ammonium hydroxide, ammonium acetate, methanol, chloroform, DIEA, DMF, DCM, THF, ethyl acetate, n-hexane, piperidine, sodium hydroxide and guanidinium hydrochloride were sourced from Merck Millipore (Australia). NMP was obtained from Applied Biosystems (USA). TFA was purchased from Auspep Pty Ltd (Australia). 33% HBr in acetic acid, tri-n-butyl phosphine, pyridine, ascorbic acid, homoserine lactone, dipyridine disulfide, tetrakis-(triphenylphosphine) palladium(0), mono-methoxytrityl chloride, iodine, TEA, DIC, TIPS, DODT, recombinant human insulin and human serum were sourced from Sigma-Aldrich

(Australia). (H-Cys-OAll)2.2Tos was obtained from Bachem (Switzerland).

2. Reversed Phase – HPLC

All RP-HPLC was done on a Waters 600 HPLC controller with a 600 pump and a 996 photodiode array detector. Mobile phases: (1) buffer A = 0.1% TFA in water, buffer B = 0.1% TFA in acetonitrile; (2) buffer A = 10 mM NH4OAc in water, pH 9; buffer B = 10 mM NH4OAc in 80% acetonitrile in water, pH 9. Analysis was performed on a Phenomenex Kinetex 5µ XB-C18, 100 Å, 50 x 4.6 mm column. Typical gradient: 10-70% buffer B over 30 minutes with a 1.5 mL/min flow rate and detection at 220 nm. Purification was carried out on a Phenomenex Kinetex 5µ XB-C18, 100 Å, 150 x 21.2 mm AXIA packed column. Typical gradient: 20-60% buffer B over 40 minutes with a 10 mL/min flow rate and detection at 230 nm. All peptides were lyophilized in a Christ freeze dryer.

155

3. Mass Spectrometry

MALDI-MS experiments were performed on a Bruker Daltonics Autoflex II TOF/TOF, using sinapinic acid as the matrix. ESI-MS experiments were performed on an Agilent 6220 Accurate-Mass TOF LC/MS.

4. Amino Acid Analysis

Amino acid analysis was performed by Auspep Pty Ltd.

5. NMR Spectroscopy

1H and 13C NMR experiments on Fmoc-γ-Br-hAla-OH and Mmt-Cys-OAll were performed on a 400 MHz Agilent MR 400 nuclear magnetic resonance spectrometer. 1H NMR experiments on the insulin analogues were performed on an 800 MHz Bruker Avance II nuclear magnetic resonance spectrometer.

6. Synthesis of Fmoc-γ-Br-hAla-OH

Homoserine lactone hydrobromide (6 mmol, 1,092 mg) was dissolved in pyridine (12 mL, 500 mM), followed by the slow addition of Fmoc-OSu (6 mmol, 2,024 mg). The solution slowly clarified and after one hour a precipitate began to form. After 4 hours, the reaction mixture was added dropwise to a vigorously stirred solution of cold water, followed by filtration of the precipitate. It was then washed with water (10 x 20 mL) and dried at the pump. 1810 mg of white solid was recovered (yield = 93%). The white solid was then treated with 33% HBr in AcOH (19mL, 300 mM) for 18 h, although the material did not fully dissolve. The solution was then added to a vigorously stirred solution of cold water, followed by filtration of the precipitate, washes with water (10 x 20 mL) and dried at the

156 pump. 2028 mg of white solid was recovered (yield = 90%). Total yield = 84%. See Figure A2.1 for analytical data.

Melting point = 156 – 162 oC. 1H NMR (400 MHz, DMSO-d6): δ = 7.88 (d, 2H, J = 7.7 Hz), 7.69 (t, 2H, J = 6.9 Hz), 7.40 (t, 2H, J = 6.9 Hz), 7.32 (t, 2H, J = 7.4 Hz), 4.30 (d, 2H, J = 7.0 Hz), 4.22 (t, 1H, J = 6.7 Hz), 4.11 (sx, 1H, 4.6 Hz), 3.60 – 3.53 (m, 1H), 3.51 – 3.43 (m, 1H), 2.25 – 2.04 (m, 2H). 13C NMR (400 MHz, DMSO-d6): δ = 173.63, 156.62, 144.21, 141.18, 128.10, 127.53, 125.68, 120.59, 66.06, 52.72, 47.10, 34.18, 31.48. Mono-isotopic molecular mass of the [M + H+]+ species via ESI-MS analysis = 404.043 Da; theory = 404.049 Da.

157

7. Synthesis of Mmt-Cys-OAll

(H-Cys-OAll)2.2Tos (4 mmol, 1,092 mg) was dissolved in THF (40 mL, 100 mM) in the presence of DIEA (3,483 µL, 20 mmol) followed by the slow addition of Mmt-Cl (4 mmol, 1,235 mg). The reaction was stirred for 4 hours, then tri-n- butylphosphine (999 µL, 4 mmol) was added, followed by H2O (720 µL, 40 mmol). The solution was stirred for a further 18 hours. Ethyl acetate (150 mL) was added to the reaction mixture which was then washed with brine (5 x 200 mL). The solution was treated with dry MgSO4, filtered and dried in vacuo. The

158 oil was purified via flash chromatography using hexane/EtOAc (0-15 EtoAc with 2.5% increments). The clear oil was recovered in good yield (1380 mg, 80% yield). See Figure A2.2 for analytical data.

Melting point = N/A (oil). 1H NMR (400 MHz, DMSO-d6): δ = 7.44 – 7.39 (m, 4H), 7.31 – 7.23 (m, 6H), 7.17 (t, 2H, J = 7.2 Hz), 6.83 (d, 2H, J = 9.0 Hz), 5.79 – 5.68 (m, 1H), 5.25 – 5.12 (m, 2H), 4.24 – 4.17 (m, 1H), 4.13 – 4.06 (m, 1H), 3.70 (s, 3H), 3.05 (d, 1H, J = 9.6 Hz), 2.53 – 2.49 (m, 1H), 2.29 (t, 1H, J = 8.7 Hz). 13C NMR (400 MHz, DMSO-d6): δ = 172.61, 158.09, 146.65, 146.50, 137.98, 132.72, 130.19, 128.70, 128.68, 128.31, 126.78, 118.39, 113.62, 70.54, 65.26, 58.23, 55.46, 28.43. Mono-isotopic molecular mass of the [M]- species via ESI-MS analysis = 432.194 Da; theory = 432.164 Da.

8. Synthesis of A11 Insulin

The A-chain was synthesized on a scale of 0.125 mmol using rink amide resin via attachment of the Fmoc-Asp-OtBu side-chain. The Cys(Dpm) residue then incorporated followed by assembly up to Ser12 using Fmoc SPPS. 2 eq. of Fmoc-γ-Br-hAla-OH was coupled via HOBt/DIC chemistry. 3 eq. of Mmt-Cys- OAll was then added in the presence of 5 eq. of DIEA in DMF and reacted for 3 hours. Cys(Acm)-Thr-Ser-Ile residues were then incorporated, followed by treatment with 3 eq. Of Pd(0) in a solution of chloroform/acetic acid/N- methylmorpholine (37/2/1) and then piperidine. The lactamization was performed using three treatments of 5 eq. of HOBt/DIC at 50 oC. N-Mmt was then cleaved with TFA/TIPS/DCM (1%/2%/97%), 10 x 2 m treatments of 5 mL), followed by Fmoc SPPS, cleavage with 15 mL of TIPS/H2O/DODT/TFA (2%/2%/1%/95%) and ether precipitation. 270 mg of crude material was recovered, which was purified in three batches using TFA buffers (pH 2) to generate 32.7 mg of purified A-chain (yield = 12%).

The B-chain was assembled via microwave assisted SPPS on a 0.1 mmol scale using preloaded Fmoc-Thr(tBu) resin, with Cys(Acm) incorporated at position

159

B7. Cleavage was performed with 15 mL of TIPS/H2O/DODT/TFA (2%/2%/1%/95%) followed by ether precipitation, resulting in the recovery of 340 mg of crude peptide. The material was purified in two batches with ammonium acetate buffers (pH 9). 54.0 mg of purified peptide was recovered (yield = 16%). 20.0 mg was then treated with 3 eq. Of DPDS (3.8 mg) in TFA (2 mL) and reacted for 1 hour, followed by ether precipitation and purification with ammonium acetate buffers (pH 9). 6.8 mg of pure peptide was recovered (yield = 33%). Total yield = 5%.

4.7 mg of A-chain and 6.8 mg of B-chain were combined in 6 M guanidinium hydrochloride (pH 8) for 90 minutes. The dimeric intermediate was then purified with ammonium acetate buffers (pH 9), resulting in a recovery of 5.0 mg (yield = 45%). All 5.0 mg of peptide was then dissolved in AcOH (3.0 mL), 60 mM HCl

(375 µL) and 10mM I2 in AcOH solution (4.0 mL) and stirred for 1 hour. The reaction was quenched with a 1 M solution of ascorbic acid and then purified with ammonium acetate buffers (pH 9) and then TFA buffers (pH 2) to remove all impurities. The final mass recovered was 1.2 mg (yield = 25%). Total folding + + yield = 11%. RP-HPLC analysis: purity > 96%. MALDI-MS analysis: [M + H ] (th.) + + = 5792.7, [M + H ] (exp.) = 5792.2. Amino acid analysis: net peptide content = 74.3%.

9. Conversion of Recombinant Human Insulin to the Trifluoroacetate Salt

30.0 mg of recombinant human insulin was purified with TFA buffers (pH 2) in order to ensure that it had the same counter-ion as A11 insulin. 26.1 mg was recovered. Amino acid analysis: net peptide content = 82.4%.

160

10. Insulin receptor binding assay

IGF-1R-negative cells over-expressing the IR-A or IR-B were generated. Cells were serum-starved for 4 h before lysis. Lysates were captured in a 96 well plate previously coated with anti-IR antibody 83-7. Approximately 500,000 fluorescent counts of europium labelled insulin were added to each well along with increasing concentrations of unlabelled competitor and incubated for 16 h at 4oC. After washing time-resolved fluorescence was measured using 340 nm excitation and 612 nm emission filters with a BMG Lab technologies Polarstar fluorometer (Mornington, Australia). Assays were performed in triplicate three times.

11. Circular Dichroism Spectroscopy

Circular dichroism spectroscopy was performed on an Applied Photophysics Chirascan Plus instrument. 25 µM aqueous solutions of both A11 and native insulins were prepared, and experiments were performed at 25 oC, across a wavelength range from 190 – 260 nm.

12. Differential Scanning Calorimetry

Differential scanning calorimetry was performed on a TA Instruments Nano DSC. 250 µM aqueous solutions of both A11 and native insulins were prepared, and experiments were performed across a temperature range from 30 – 90 oC.

13. Human Serum Stability Assay

475 µL of human serum was transferred to a 1.5 mL tube and immersed in a water bath set to 37 oC for 10 minutes. 50 µg of peptide was dissolved in 25 µL of water and then mixed in with the serum. 90 µL aliquots at five time points (0,

161

60, 120, 240 and 480 minutes) were pipetted into a second 1.5 mL tube containing 300 µL of a cooled solution of 1% formic acid in 90% aqueous acetonitrile. The tubes were cooled on ice for 45 minutes, followed by centrifugation at 12,000 x g at 4 oC for 15 minutes. 250 µL of the supernatant was transferred to a third 1.5 mL tube containing 100 µL of water and then frozen and lyophilized. Each sample was then dissolved in 0.1% TFA in 10% aqueous acetonitrile and analyzed via RP-HPLC on a C4 column. The area under the peak was determined for each time point and the amount of degradation was expressed as a percentage of peptide at time zero. Both the A11 and native insulins were assayed in triplicate.

14. Thermal Stability Assay

70 µg of peptide was dissolved in 700 µL of water and transferred to a 1.5 mL tube and immersed in a water bath set to 60 oC. 110 µL aliquots at five time points (0, 1, 2, 3 and 4 days) were pipetted into a second 1.5 mL tube and frozen at -20 oC. Once the time course was complete, each sample was then analyzed via RP-HPLC on a C18 column. The area under the peak was determined for each time point and the amount of degradation was expressed as a percentage of peptide at time zero. Both the A11 and native insulins were assayed in triplicate.

162

163

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Karas, John Andrew

Title: Enhanced synthetic methods and stabilization of insulin-like peptides

Date: 2015

Persistent Link: http://hdl.handle.net/11343/59468

File Description: Enhanced Synthetic Methods and Stabilization of Insulin-like Peptides