Synthesis of β-turn and pyridine based peptidomimetics

(1)

Synthesis of β-Turn and Pyridine Based

Peptidomimetics

David Blomberg

(2)

Department of Chemistry, Organic Chemistry Umeå University

SE-901 87 Umeå

Copyright  2007 by David Blomberg ISBN 978-91-7264-305-5

(3)

1. List of Papers

I David Blomberg, Mattias Hedenström, Paul Kreye, Ingmar Sethson, Kay Brickmann and Jan Kihlberg; Synthesis and conformational studies of a β-turn mimetic incorporated in Leu-enkephalin. J. Org. Chem., 2004, 69, 3500-3508.

II David Blomberg, Paul Kreye, Kay Brickmann, Chris Fowler and Jan Kihlberg; Synthesis and biological evaluation of leucine enkephalin turn mimetics. Org. Biomol. Chem., 2006, 4, 416-423.

III David Blomberg, Kay Brickmann and Jan Kihlberg; Synthesis of a β-strand mimetic based on a pyridine scaffold. Tetrahedron,

2006, 62, 10937-10944.

IV David Blomberg, Tomas Fex, Yafeng Xue, Kay Brickmann and Jan Kihlberg; Design, synthesis and biological evaluation of thrombin inhibititors based on a pyridine scaffold. Submitted.

V David Blomberg, Tomas Fex, Kay Brickmann and Jan Kihlberg; Design, synthesis and biological evaluation of thrombin inhibitors lacking a strong basic functionality in P1. Manuscript.

(6)

(7)

2. List of Abbreviations

Bn benzyl

Boc tert-butoxycarbonyl

BSA bovine serum albumin

Cbz benzyloxycarbonyl

DAMGO [D-Ala2,N-Me-Phe4,Gly5-ol]-enkephalin DBU 1,8-diazabicyclo[5.4.0]undec-7-ene DCC dicyclohexyl carbodiimide

DCE 1,2-dichloroethane

DIBAL diisobutylaluminium hydride DIC N,N’-diisopropyl carbodiimid DIPEA diisopropylethylamine DMAP N,N-dimethylaminopyridine DPDPE [3H] [D-Pen2, D-Pen5]enkephalin DTI direct thrombin inhibitor

EWG electron withdrawing group Fmoc 9-fluorenylmethyloxycarbonyl GPCR G protein-coupled receptor HATU O-(7-azabenzotriazole-1-yl)-N, N,N’N’-tetramethyluronium hexafluorophosphate HMDS hexamethyldisilazane HOAt 1-hydroxy-7-azabenzotriazole HOBt N-hydroxybenzotriazole

K-selectride potassium tri-sec-butylborohydride LCMS liquid chromatography mass spectrometry LHRH luthenizing hormone releasing hormone N,O-DMHA N,O-dimethylhydroxylamine

NMO N-methyl morpholine N-oxide NMR nuclear magnetic resonance NOE nuclear overhauser enhancement

(8)

TBAF tetrabutylammonium fluoride TBDMS tert-butyldimethyl silyl

TEA triethylamine

TMP 2,4,6-trimethylpyridine TFA trifluoro acetic acid TFAA trifluoro acetic anhydride TFE 2,2,2-trifluoroethanol

TMS trimethylsilyl

Tris tris hydroxymethylaminoethane

UHP urea hydrogen peroxide

(9)

3. Introduction

3.1 Structure and function of peptides and proteins

On a molecular level proteins are built up by small residues, amino acids, that are connected via amide bonds to form chains (Figure 3.1). Shorter amino acid sequences, usually containing 2−50 amino acid residues, are defined as peptides, while longer chains are defined as proteins. There are 20 naturally occurring amino acids with a variety of characteristics in the side chains (R1−R4) to provide a wide scope of peptides and proteins with enormous variation in properties and function. Biologically active peptides often act as hormones or transmittor substances and regulate several vital physiological functions, such as secretion of growth hormone (somatostatin), intake of food (neurotensin), sexual function (melanocortin-4) and blood pressure (angiotensin, vasopressin).1-3 Enzymes, transporters over membranes and receptors are examples of important physiological functions filled by proteins. N H H N N H H N R1 O R2 O R3 O R4 Amino acid residue Amide bond

Figure 3.1 A short amino acid sequence with an amino acid residue and an amide

bond highlighted. R1−R4 represent amino acid side chains.

The amino acid sequence is referred to as the primary structure of peptides and proteins. Peptides and proteins are usually arranged in more stable secondary structures such as α-helices, β-turns and β-strands.4

(10)

Figure 3.2 A schematic picture of a GPCR with α-helices as transmembrane

elements.

The α-helix is a common structural elements in proteins, for example in the transmembrane regions of GPCR’s (Figure 3.2). The α-helix is stabilized by intramolecular hydrogen bonds between amino acid residues i+4 and i, to form a right handed helical structure. Proline and glycine are rare in α-helices due to the lack of ability to donate a hydrogen bond and high flexibility, respectively.

β-Turns occur frequently in proteins and are sequences where the peptide backbone reverses its overall direction over four amino acids, i to i+3 (Figure 3.3). There are several different types of β-turns defined by the dihedral angels, φ and Ψ, of amino acids i+1 and i+2. A β-turn is generally

stabilized by an intramolecular hydrogen bond to form a pseudo ten membered ring and is believed to be of great importance for the recognition and activity of peptides and proteins.5-8

N HN O H HN O O ! 1 1 "

Figure 3.3 A β-turn with the dihedral angels φ and Ψ shown for residue i+1. The

structure only contains Ala for clarity.

A β-strand is an amino acid sequence with the peptide backbone in an extended conformation. It is rarely found as a monomer, but more often as dimers or in tertiary structures called β-sheets, stabilized by hydrogen bonds and hydrophobic interactions (Figure 3.4). The correct overall orientation of the secondary structure elements, i.e. the tertiary structure, of a peptide or a protein is crucial both to get stability and proper biological functions.9,10

(11)

H N N H H N N H O O O O H N OC H N NH O O N H O H N O N H O CO HN H N H N O N H H N O O

Figure 3.4 A β-strand (only containing Ala for clarity) assembled to form a β-sheet.

The structure is stabilized by intramolecular hydrogen bonds and hydrophobic interactions.

Despite the versatile and interesting function of peptides in biological systems, metabolic instability and poor bioavailability make them ineffective as orally administered drugs. A few peptides are, however, used as drugs today, such as insulin to control blood sugar levels and oxytocin to induce uterus contractions. Due to the problems with oral administration, insulin and oxytocin have to be given subcutaneously or intravenously. To avoid the problems associated with oral administration, peptides can serve as templates to provide a pharmacophore model in the development of compounds exerting the same action as the native peptide, but with improved pharmacokinetic properties.

3.2 Peptidomimetics

The unfavorable pharmacokinetic properties associated with peptides when used as orally administered drugs can, in principle, be avoided by development of peptidomimetics. The general strategy when preparing peptidomimetics is to replace segments related to undesired properties with non-peptidic structures, while attempting to maintain the ability to elicit the same or improved biological response as the native peptide.11,12

Peptidomimetics can be divided into three different classes.3_{The first}

class is characterized by backbone changes, such as incorporation of amide bond isosteres and turn mimetics. The vast literature concerning peptidomimetics of class I including stabilized turn mimetics represented as bicycles,13-19_aromatics20,21_{and cyclic compounds.}22_{The second class of}

peptidomimetics are referred to as ligands exerting the same biological response as the native peptide ligand without any obvious structural

(12)

3.2.1 Development of a peptidomimetic drug - Exanta

Today the major drugs used as anticoagulants are warfarin (Waran) and heparin, which inhibit the blood coagulation cascade by two different pathways (Figure 3.5 and Chapter 7.1). Warfarin is a vitamin K antagonist that disturbs formation of the coagulation factors in the coagulation cascade. Heparin induces a conformational change in endogenous antithrombin which increases its activity and thereby inhibits thrombin. Warfarin is the only anticoagulantia that is given orally, but suffers from a narrow therapeutic window and is known to have severe drug-drug and drug-food interactions. To avoid over- or under-treatment, careful dosing and monitoring of the plasma concentration has to be performed regularly. The major drawbacks with heparin is the high molecular weight (∼1500-20000 g/mol) together with highly polar functional groups, which demands subcutaneous administration. Since both current therapies are associated with problems, a new oral anticoagulantia with a wide therapeutic window is highly desirable.

O O OH O Warfarin O O O O O O3S O O3S O2C O HO O O3S O3S O O O2C O Heparin O AcHN NH HO _OH HO HO

Figure 3.5 The structures of warfarin and heparin; the two most common

anticoagulantia.

In the mid 80’s a research team at AstraZeneca R&D in Mölndal initiated a project directed towards an orally administered reversible direct thrombin inhibitor (DTI).27_{A DTI exerts its action by blocking the active site of the}

enzyme thrombin, that is the last factor in the blood coagulation cascade. The first approach was to mimic a pentapeptide (D-Phe-Val-Arg-Gly-Pro) that had been designed based on a proposed conformation of fibrinogen in the active site of thrombin. The labile amide bonds were replaced by proteolytically more stable ketomethylene isosteres. Acceptable potencies were obtained in this series of compounds, but the pharmacokinetic profile was unsatisfactory. In the early 90’s AstraZeneca gained access to the crystal structure of thrombin and a new approach using computer modeling was started. This new approach eventually revealed melagatran as a potential

(13)

DTI with a molecular weight less than 500 g/mol (Figure 3.6). In vitro and in vivo (iv administration) confirmed melagatran as a reversible DTI with a much wider therapeutic window than warfarin.

NH2 NH N H O N O H N HO O Melagatran H N NH N H O N O H N O O OH Ximelagatran

Figure 3.6 The thrombin inhibitor Exanta; structures of the active substance

melagatran and the prodrug ximelagatran.

The secondary amine, the carboxylic acid and the benzamidine moiety are all charged to a high extent at physiological pH (7.2-7.4). Not surprisingly the highly polar substance suffered from poor membrane permeability and extremely low bioavailability. Initial attempts to overcome the permeability issue were made by exploring different formulations, but these efforts were unfruitful. Instead an approach to make melagatran more lipophilic by using a dual prodrug, both at the amidine and the carboxylic acid, was investigated. The prodrug strategy resulted in ximelagatran which had good permeability and is converted to melagatran in vivo. Ximelagatran stood up to all project goals as an orally administered DTI and the drug Exanta was released to the market in 2003. The research behind Exanta is an interesting story of the development of a peptidomimetic that finally became a drug and reached the market.

(14)

4. Objectives of the thesis

The overall objective of this study was to develop and apply synthetic organic chemistry in the synthesis of peptidomimetics with the focus set on mimicry of β-turn and β-strand peptide secondary structures. In order to reach this objective the following approaches were taken.

• Stabilized turn mimetics were to be synthesized in order to gain further understanding of the bioactive conformation of the endogenous endorphin Leu-enkephalin (H2

N-Tyr-Gly-Gly-Phe-Leu-OH).

• A scaffold based strategy towards β-strand mimetics was to be developed. The scaffold should promote an extended conformation and allow most of the hydrogen bonding capacity to be maintained. • Experience acquired during development of the chemistry towards

the turn and strand mimetics, was to be applied to suitable medicinal chemistry projects.

(15)

5. Peptidomimetics of Leu-enkephalin

5.1 Biological action and conformation of

Leu-enkephalin

Leu-enkephalin (Tyr-Gly-Gly-Phe-Leu) is an endogenous peptide which was discovered in the mid 1970s (Figure 5.1).28_{The peptide is generated from}

proenkephalin and released in the brain, where it acts as an agonist at the µ, δ, and κ opiate receptors The opiate receptors belongs to the rhodopsin subclass within the superfamily of G protein coupled receptors (GPCR’s), which are characterized by seven transmembrane α-helices.29,30 The Leu-enkephalin pentapeptide interacts with the receptors and a cascade mechanism that eventually results in relief of pain is initiated.31_{Soon after}

the discovery of Leu-enkephalin a crystal structure of the peptide was reported to contain a 1−4 β-turn conformation in the solid state.32_An

intramolecular hydrogen bond between the carbonyl oxygen in Tyr (i) and the amide nitrogen in Phe (i+3) stabilizes the β-turn, thus forming a pseudo ten membered ring.

HN OHN H N O O H2N OH NH O O HO Leu-Enkephalin

Figure 5.1 Leu-enkephalin drawn as a pseudo ten membered ring stabilized by an

(16)

enkephalin in the pseudo ten membered ring observed in the crystalline form.33

HO O OH

N Me

Figure 5.2 The rigid morphine.

5.2 Design of Leu-enkephalin peptidomimetics

As mentioned above several reports, propose that Leu-enkephalin could be stabilized by an intramolecular hydrogen bond, either forming a 1−4 or a 2−5 β-turn, in the bioactive conformation.34_{In addition a pharmacophore}

model in which, the charged N-terminal amine, the aromatic residue of Tyr, and the hydrofobic residues of Phe and Leu are key interaction points should be considered when developing peptidomimetics.35-39_{In this study, the}

assumption that the crystal structure of Leu-enkephalin resembles the bioactive conformation was made. In order to stabilize Leu-enkephalin in a 1−4 β-turn conformation, the intramolecular hydrogen bond that forms the pseudo ten membered ring was replaced by an ethylene bridge, affording a stable ten membered ring as in 1 (Figure 5.3).40_{The amide bond between}

Tyr (i) and Gly (i + 1) was replaced with an ether linkage which had been shown to retain biological activity in the linear peptidomimetic 3.41_The

methylene ether isostere was employed in order to make the synthesis more feasible while simultaneously striving to ensure biological acceptance for the inserted features. Additionally, to explore the importance of the ring size and the positions of the key features suggested by the pharmacophore model, a Leu-enkephalin analogue 2, containing a seven membered ring, was designed. Linear peptide analogues 3 and 4 were also synthesized as controls to probe the effect of cyclization.

(17)

O HN N O O H2N OH NH O O HO 1 H2N N O O H N O O OH HO H2N O H N OH O H2N O H N OH O 2 4 3 N H O N H O H N O OH O OH O

Figure 5.3 The ten membered β-turn mimetic 1, and the seven membered β-turn

mimetic 2, inserted in Leu-enkephalin. Linear analogues 3 and 4. All compounds contain an identical amide isostere.

The designed β-turn mimetic was also aimed to be inserted in other bioactive peptides, and a stabilized β-turn mimetic corresponding to residues of LHRH has been prepared by the same synthetic strategy.25

5.3 Synthesis of peptidomimetics incorporated in

Leu-enkephalin

5.3.1 Synthesis of a ten membered β-turn mimetic

The strategy behind the synthesis was to introduce all structural features followed by a ring closure via a macrolactamization to afford mimetic 1. A retrosynthetic analysis (Figure 5.4) revealed that 1 could be synthesized from aldehyde I, dipeptide II and glycine III, all suitably protected. Dipeptide II and aldehyde I were to be connected by a reductive amination, followed by an amide bond formation between glycine III and the secondary amine formed in the reductive amination. With all residues in place, ring closure to the desired ten membered ring was planned to be achieved via an amide bond formation. Building blocks II and III were commercially available, but aldehyde I had to be synthesized.

(18)

HN N O H2N OH O NH O O O HO NH2 H2N OH O NH O O O HO OH O I II 1 O III HO H2N

Figure 5.4 Retrosynthetic analysis of the ten membered β-turn mimetic 1.

The synthesis started by preparation of Weinreb’s amide 5 (81%) from Cbz-Tyr(OtBu)-OH using iso-butylchloroformate, N-methyl morpholine and N,O-dimethylhydroxylamine in CH2Cl2 (Scheme 5.1).42 Weinreb’s amide 5

was then treated with allylmagnesium bromide in THF to afford ketone 6 (74%).43 OtBu NHCbz O OH OtBu CbzHN O N Me OMe OtBu NHCbz O 81% 74% NO-DMHA AllylMgBr 6 5

Scheme 5.1 Preparation of homoallyl ketone 6.

Different reducing agents were then applied to ketone 6. Chelating reducing agents such as DIBAL or Zn(BH4)2 gave anti aminoalcohol 8 as the major

product, while the non-chelating K-selectride favored syn aminoalcohol 7 with a selectivity of ∼ 6:1 (Scheme 5.2).44,45 When using K-selectride as reducing agent, oxazolidinone 9 (14%) was also formed from 7, but the other diastereomeric oxazolidinone was not observed. For the chelating reducing agents no formation of oxazolidinone was observed, probably due to decreased nucleophilicity of the generated alcohol when complexed by Zn or Al.

(19)

OtBu NHCbz O 6 K-selectride OtBu NHCbz OH OtBu NHCbz OH + + HN O H O OtBu 7, 36% 9, 14% 8, 8% H

Scheme 5.2 The stereoselective reduction of ketone 6.

The configuration of alcohols 7 and 8 was determined after conversion to the corresponding oxazolidinones (Scheme 5.3). Thus, 8 and 7 were treated with aqueous potassium hydroxide in a mixture of methanol and THF, and the NOE-spectra of the generated oxazolidinones were compared.40

OtBu NHCbz OH OtBu NHCbz OH HN O O OtBu 7 9 8 HN O O OtBu 7.5 M KOH (aq) MeOH THF 7.5 M KOH (aq) MeOH THF 10 H-1 H-1 H-2 H-2

Scheme 5.3 Conversion of alcohols 7 and 8 to their corresponding oxazolidinones.

Irradiation of H-2 in 9 gave a 5% NOE (selective) for H-1, whereas a 14% enhancement was obtained for the diastereomeric oxazolidinone 10. This established an anti relationship between H-1 and H-2 in 9 and a syn relationship in 10 (Scheme 5.3). Further into the synthetic sequence it was discovered that only the syn aminoalcohol 7 could be converted to the desired β-turn mimetic. Therefore, K-selectride was chosen as reducing agent to raise the yield of syn aminoalcohol 7 in the transformation of 6.

(20)

with triflic anhydride, sodium azide and CuSO4 to afford azidoalcohol 12 (85%).46,47 7 or 9 KOH H2O/EtOH NH2 OH OtBu 80% triflic anhydride NaN3 CuSO4 N3 OH OtBu 85% QHSO4 NaOH (aq) t-Bu bromoacetate N3 O OtBu OtBu O OsO4 NMO 73%, from 12 11 N3 O OtBu OtBu O OH OH N3 O OtBu OtBu O O 12 15 Pb(OAc)4 Na2CO3 85% 13 14

Scheme 5.4 Synthetic procedure to aldehyde 15.

Azidoalcohol 12 was alkylated with α-bromo tert-butylacetate under biphasic conditions, with QHSO4 as phase transfer catalyst, to afford 13.48

Compound 13 was found to be unstable both during purification on silica gel and under storage. Crude 13 was therefore directly oxidized to the stable diol

14 (73%, from 12) using OsO4 and NMO in a solvent mixture containing

acetone, THF and H2O. The synthetic sequence was continued by further

oxidation of the diastereomeric mixture of diol 14 to aldehyde 15 (85%) using Pb(OAc)4 and Na2CO3 in benzene. Alkylation of the diastereomeric

azidoalcohol, generated by same synthetic sequence as 12, but using 8 as starting material, proved unsuccessful when applying the same conditions as above. Most likely this was due to rapid decomposition of the corresponding diastereomer to 13 during synthesis or on purification.

Next, H2N-Phe-Leu-OH was converted to H2N-Phe-Leu-OMe (16) in almost

quantitative yield by treatment with TMSCl in MeOH. The following reductive amination was performed in DCE with Na(OAc)3BH as reducing

agent under basic conditions to afford 17 (82%) with all motifs of the β-turn mimetic present except for Gly (i+2) (Scheme 5.5).49_{The remaining glycine}

was introduced via an amide bond formation promoted by HATU and DIPEA in DMF to obtain 18 (85%).50,51

(21)

N3 O OtBu OtBu O 15 + 16 NH H N O MeO O Na(OAc)3BH TEA DCE 82% N3 O OtBu OtBu O N H N O MeO O HATU DIPEA DMF Fmoc-Gly-OH 85% O NHFmoc formic acid 99% N3 O OH OH O N H N O MeO O O NHFmoc 17 18 19 N3 O OH OPfp O N H N O MeO O O NHFmoc DCC PfpOH EtOAc N3 O OH N H N O MeO O O HN O 20 21 1 R = OH NH2 O OH N H N O R O O HN O DBU dioxan reflux 56%, from 19 SnCl2 PhSH TEA 57%, from 21 22 R = OMe LiOH

Scheme 5.5 Completing the synthesis of β-turn mimetic 1.

The phenolic tert-butyl ether and the tert-butyl ester in 18 were simultaneously removed to give 19 (99%) using formic acid, leaving the methyl ester unaffected. Activation of the carboxylic acid to afford pentafluorophenyl ester 20 was accomplished by using pentafluorophenol and DCC in EtOAc.52_{Despite purification on silica gel 20 was contaminated}

with dicyclohexyl urea, but the sequence was continued without further purification. The activated acid 20 was dissolved in dioxane (35 mg/mL) and added to refluxing dioxane (0.15 mL/mg of 20) containing DBU via a syringe pump over a period of 12 h. After the addition was completed, the reaction was refluxed for another 20 minutes to give ringclosed 21 (56%, from 19). Attempts to complete the critical lactamization starting with Fmoc deprotection, followed by using amide bond coupling reagents (DIC, HATU) proved less successful. Finally, reduction of the azide53 and hydrolysis of the methyl ester were achieved in a two step procedure. First reduction of the azide using SnCl2, PhSH and triethylamine in THF, followed by purification

(22)

5.3.2 Conformational studies of the ten membered β-turn

mimetic using NMR spectroscopy

Several attempts were made to obtain high quality spectral data of β-turn mimetic 1 in water, DMSO-d6/water, TFE-d3/water as well as MeOH-d4.

However, spectral data with sufficient quality for assignment of 1 could only be achieved in aqueous solution. In water broad peaks and cross peaks with both negative and positive signs in NOESY indicated a slow exchange between several conformations. Therefore compound 21, also containing the ten membered β-turn mimetic, was used as a model for mimetic 1 in the conformational studies. To our delight, for 21 which has both the C- and N-termini protected, spectral data with sufficient quality for conformational determination was obtained in MeOH-d3 at – 70 °C (Figure 5.5).

Figure 5.5 a) A superimposition of the ten conformers with lowest energy generated

from the NMR study. b) An overlap between the crystal structure of Leu-enkephalin (dark) and one low energy conformer of 21 (bright). c) An overlap of the β-turn mimetic in 21 (dark) and an idealized type II β-turn (bright).

The extracted conformations were superimposed on different idealized β-turns, which revealed that the highest similarity for an ensemble of low energy conformations was obtained with a type II β-turn (backbone rmsd = 0.43 Å). The ensemble of low energy conformers was also compared with the ten membered ring of crystalline Leu-enkephalin (rmsd = 0.55 Å for the heavy atoms). The only larger deviation of the mimetic from an ideal type II β-turn and Leu-enkephalin in the solid state, was the side chain orientation of the second amino acid (i+1), where a D-amino acid would give an

improved fit. The fact that the second amino acid in Leu-enkephalin is glycine, implies that this deviation should not affect binding to the opiate receptors.

(23)

5.3.3 Synthesis of a seven membered β-turn mimetic on solid

phase

The seven membered β-turn mimetic 2 constitutes a truncated form of Leu-enkephalin, lacking one of the two glycine residues. The synthetic procedure is mainly based on the strategy developed for 1 (see section 5.3.1) but the assembly of the building blocks was performed on solid phase.

NHBoc O O H N O O O NH2 2. DIC HOBt Fmoc-Phe-OH 1. TFA/CH2Cl2 3. Piperidine DMF N3 NH O H N O O O tBuO O OtBu 15 Na(OAc)3BH TEA TFA/CH2Cl2 N3 NH O H N O O O OH O OH N3 NH O H N O O O OH O OPfp PfpOH DIC 23 24 ₂₅ ₂₆

Scheme 5.6 Synthesis of activated acid 26, ready for the ring closing reaction.

The main advantage when performing some of the synthetic steps toward 2 on solid phase was that the crucial lactamization was conducted when attached to solid phase, thereby eliminating the risk for oligomerization. First, Boc-protected Leu on solid phase (Tentagel, PAM linker) was elongated with Phe using the standard Fmoc procedure (Scheme 5.6).54_The

solid phase bound dipeptide 23 was reacted with aldehyde 15 in a reductive amination reaction using Na(OAc)3BH and triethylamine in DCE to afford

secondary amine 24.49_{Next, the phenolic tert-butyl ether and the tert-butyl}

ester was cleaved using TFA in CH2Cl2 to generate carboxylic acid 25,

which was activated as a pentafluorophenol ester 26 using DIC and pentafluorophenol.52_{Ring closure was accomplished in refluxing dioxane}

(24)

N3 N O O H N O O O HO TEA dioxane reflux N3 N O O H N O O OMe HO H2N N O O H N O O R HO NaOMe MeOH SnCl2 TEA PhSH 25%, over solid phase synthesis 26 27 28 2 R = OH 56%, from 28 29 R = OMe LiOH

Scheme 5.7 Completing the synthesis of β-turn mimetic 2.

Ringclosed 27 was released from the solid phase using NaOMe in methanol to afford protected turn mimetic 28 (25%, over 8 steps on solid phase). β-Turn mimetic 28 was then deprotected in a two step procedure starting with reduction of the azide53_{to the corresponding amine 29 using SnCl}

2, TEA and

PhSH in THF. Hydrolysis of the methyl ester and purification by reversed phase HPLC then afforded target β-turn mimetic 2 (56%, from 28).

5.3.4 Synthesis of linear Leu-enkephalin analogues

Fmoc-Tyr(OtBu)-OH was initially reduced to the corresponding alcohol 30 (90%) by activating the acid with iso-butyl chloroformate followed by reduction with NaBH4 (Scheme 5.8).55 The Fmoc-protected amine was then

liberated using morpholine to afford aminoalcohol 31 (90%). An azidotransfer reaction using triflic anhydride and NaN3 in a biphasic system

then afforded azidoalcohol 32 (72%).46,47

OtBu NHFmoc OH O OtBu NHFmoc OH iBu-Chloroformate NaBH4 90% OtBu NH2 OH Morpholine OtBu N3 OH 90% 72% Tf2O NaN3 30 31 32

Scheme 5.8 The synthesis of azidoalcohol 32 from the corresponding

Fmoc-protected amino acid.

Next, alcohol 32 was alkylated with ethyl bromoacetate using KH and QI in THF at 0 °C to afford 33 (89%, Scheme 5.9). Hydrolysis using NaOH in

(25)

EtOH gave 34 (80%) suitably protected for incorporation on solid phase via formation of an amide bond. Direct substitution of 32 with bromo acetic acid was less effective, hence the synthetic route via ester 33 was preferred.

OtBu N3 OH 32 OtBu N3 O OEt O OtBu N3 O OH O 33 34 Et-bromoacetate KH QI 89% 80% EtOH NaOH

Scheme 5.9 Synthesis of the dipeptide isostere 34 suitably protected for solid phase

synthesis.

Solid phase bound peptides 35 and 36 were synthesized according to the standard Fmoc procedure starting from solid phase bound Fmoc-Leu (Scheme 5.10).54_{Incorporation of carboxylic acid 34 to afford solid phase}

bound peptidomimetics 37 (from 35) or 38 (from 36), was achieved via an amide bond formation using HATU and DIPEA in DMF (Scheme 5.11).50,51

FmocHN O O H N O O O H2N H N O O O N H O H2N 35 36

Scheme 5.10 Preparation of solid phase bound peptides.

Reduction of azides 37 and 38 was performed using SnCl2, TEA and PhSH

in THF to afford the corresponding amines 39 and 40.53_{A cleavage mixture}

containing TFA, H2O, ethanedithiol and thioanisol was added to deprotect

and release the peptidomimetics from the solid phase to afford 4 (62%, over the solid phase synthesis) and 3 (70%, over the solid phase synthesis) after purification on reversed phase HPLC.

(26)

H N O O O N H H N O O O N H O H N O O O O N3 OtBu N3 OtBu H N OH O O N H O H N O O H2N OH H N OH O O N H O O H2N OH 35 36 H N O O O N H O H N O O H2N OtBu H N O O O N H O O H2N OtBu 37 38 4 3 39 40 34 HATU DIPEA CH2Cl2 34 HATU DIPEA CH2Cl2 SnCl2 TEA PhSH THF SnCl2 TEA PhSH THF H2O TFA scavangers H2O TFA scavangers 62% over solid phase synthesis 70% over solid phase synthesis

Scheme 5.11 Completing the synthesis of peptidomimetics 4 and 3.

5.4 Biological evaluation

5.4.1 Opioid receptor binding assay

Leu-enkephalin mimetics 1, 2, 3 and 4 (Figure 5.6) were evaluated as specific binders at the µ and δ opiate receptors. Membrane bound opiate receptors, prepared as previously described from rat brain without cerebellum, were used in the binding assay.56_{All compounds were tested in}

triplicates at each concentration using three different rat brain homogenates (Figure 5.7 and Figure 5.8).

(27)

O HN N O O H2N OH NH O O HO 1 H2N N O O H N O O OH HO H2N O H N OH O H2N O H N OH O 2 4 3 N H O N H O H N O OH O OH O

Figure 5.6 The four synthesized peptidomimetics evaluated in an opioid binding

assay.

Receptor specific binding was measured using [3_{H] DAMGO}37_{and [}3_H]

DPDPE36_{as radioligands for the µ- and δ- receptors, respectively. In order to}

measure the receptor specific binding the membrane bound radioactivity was counted and then corrected for unspecific binding using naloxone. The reference ligands DSLET (D-Ser-Leu-enkephalin-Thr) and Leu-enkephalin were also included in the binding study and treated in the same way as the peptidomimetics for comparison. To prevent proteolysis of the amide bonds in the different Leu-enkephalin analogues, the protease inhibitor bacitracin was used in all experiments.

5.4.2 Binding to µ- and δ- opioid receptors

Among the tested peptidomimetics, linear analogue 3 showed the highest affinity for both receptors (IC50 = 14 nM at µ, 1.3 nM at δ) (Figure 5.7,

Figure 5.8, Table 5.1 and Table 5.2). Interestingly, 3 bound with higher affinity to both receptors than the two positive controls Leu-enkephalin and DSLET. The affinity of 3 confirms that the ether linkage used as amide bond isostere in our study is well tolerated without reduction of binding affinity as reported previously.41

(28)

Figure 5.7 Competitive inhibition at the µ-receptor for peptidomimetics 1 (), 2 (), 3 () , 4 () and the two known ligands DSLET, () and Leu-enkephalin (). For clarity only a few error bars are shown in the figure.

Table 5.1 Calculated IC50 values and Hill slopes for the µ-opiate receptor.

Compound pIC50 IC50 (nM) nh 1 <6 >1000 - 2 6.13 ± 0.12 740 0.90 ± 0.24 3 7.86 ± 0.09 14 0.80 ± 0.12 4 <5.5 >1000 - Leu-enkephalin 6.81 ± 0.12 160 0.66 ± 0.12 DSLET 7.41 ± 0.07 39 0.74 ± 0.01

Ten membered ring 1 showed no affinity (IC50 > 1000 nM) at any of the two

receptors. The lack of affinity is not due to the amide bond isoster of 1 since this structural fragment also is present in 3. Instead the rigidity imposed by the cyclization may prevent 1 from adopting the required conformation for binding, implying that a 1−4 type II β-turn is not the biologically active conformation for Leu-enkephalin. Other reasons for the lack of affinity could be the steric demand of the ethylene bridge in 1 that might hinder accessibility to the receptors, or the loss of a potential hydrogen bond from the Phe NH. The truncated linear analogue 4 showed no affinity for the µ-receptor (IC50 > 1000 nM) and only modest binding to the δ-receptor (IC50 =

370 nM). This suggests that the decreased length may prevent 4 from reaching critical interaction points in the binding sites of the receptors.

(29)

Figure 5.8 Competitive inhibition at the δ-receptor for peptidomimetics 1 (), 2 (), 3 () , 4 () and the two known ligands DSLET, () and Leu-enkephalin (). For clarity only a few error bars are shown in the figure.

Table 5.2 Calculated IC50 values and Hill slopes for the δ-opiate receptor.

Compound pIC50 IC50 (nM) nh 1 <6 >1000 - 2 6.79 ± 0.12 160 0.98 ± 0.26 3 8.88 ± 0.08 1.3 0.92 ± 0.14 4 6.07 ± 0.14 370 1.03 ± 0.37 Leu-enkephalin 8.44 ± 0.15 3.6 0.67 ± 0.15 DSLET 8.83 ± 0.16 1.5 0.58 ± 0.14

Seven membered ring 2 showed affinities superior to linear analogue 4 for both the µ-receptor (IC50 = 740 nM) and at the δ-receptor (IC50 = 160 nM).

The significantly increased binding affinities displayed by 2 as compared to the flexible linear analogue 4 indicates that Leu-enkephalin probably binds to both receptor subtypes in a rigid turn conformation.

5.5 Summary

Several reports claim that the flexible Leu-enkephalin (Tyr-Gly-Gly-Phe-Leu) binds to the opiate receptors in a stabilized manner forming either a 1−4 or a 2−5 β-turn. In an attempt to obtain more information concerning the bioactive conformation of Leu-enkephalin, four peptidomimetics were synthesized and evaluated in an opioid receptor binding assay. Two of the peptidomimetics were conformationally restricted by replacing the proposed

(30)

cyclization linear analogues of the two cyclized β-turn mimetics, containing the amide isostere between Tyr (i) and Gly (i+1), were also synthesized. All four compounds were tested in an opioide binding assay. This revealed that Leu-enkephalin probably binds to the µ- and δ- receptor subtypes in a more rigid turn conformation, but not in a 1−4 β-turn as we suggested in the original design.

(31)

6. β-Strand peptidomimetics

6.1 β-Strands

β-Sheets account for more than 30% of all protein structures and are built up by β-strands which have an extended conformation. The β-strand is rarely found as a monomer, instead β-strands assemble either in a parallel or an anti-parallel fashion to form β-sheets stabilized by a hydrogen bonding network (Figure 6.1).9,10 N N N H N N H N N N H N N H H N O O N N O O O O O H H H H H H H H O N H H N N H O O O O O O O O

Figure 6.1 Extended peptide chains, β-strands, dimerizes to form β-sheets. The

peptide backbone of the two involved β-strands can run in the same direction (parallel, to the left) or in opposite direction (antiparallel, to the right).

Often β-sheets serve as scaffolds, which position the side chains of proteins in positions required to elicit a biological activity. Sheets, or rather the β-strand, is recognized in several protein-protein interactions, for example by proteolytic enzymes, major histocompatibility complex (MHC) molecules, farnesyl transferases and SRC kinases. β-Strands are also involved in protein-DNA interactions in normal gene regulation by the met repressor that requires a dimerization through β-sheet domains. β-Strands are also involved in development of neurological diseases, such as Alzheimer’s and Parkinson’s disease that are associated with aggregation of proteins to form insoluble β-sheet structures. In the field of medicinal chemistry, the β-strand is regarded as a fundamental structural element for development of

(32)

retained in applications toward proteases (Figure 6.2).57-60_{Also the 1.2}

dihydro-3(6H)-pyridinone has successfully been used as a key building block in the synthesis of mimetics (@-tides) of β-strands containing up to of 13 amino acid residues.61-63_{A β-strand mimetic must maintain the capability}

to mimic the activity of the original peptide, while at the same time provide improved pharmacokinetic properties as compared to the native peptide.

N O N H R H N O O N H R N O HN O HN O HN O R R R

Figure 6.2 Two examples of β-strand mimetics. @−Tides (left) and

polypyrrolinones (right).

The maintenance of the structural features of the original peptide, combined with introduction of surrogates for amide bonds susceptible to proteolysis, is crucial in the design of a β-strand mimetic. Additionally, the structural elements of a β-strand mimetic should promote an extended backbone conformation and also retain the capability to accept and donate hydrogen bonds to facilitate aggregation into β-sheets.

6.2 Design and retrosynthetic analysis of a β-strand

mimetic

In this study a synthetic procedure to β-strand mimetic Va, based on a pyridine scaffold, was developed (Figure 6.3). The pyridine ring replaces the central amino acid within an amino acid sequence containing three residues. The NH attached at position 2 of the pyridine ring, together with the pyridine nitrogen atom mimics the second amide bond in the peptide chain, while the keto functionality attached at position 4 of the pyridine ring mimics the first amide bond of the peptide sequence.64_{The two amide bonds are thereby}

replaced with proteolytically stable functionalities and the hydrogen bonding potential, crucial for β-sheet formation, is partly maintained. Previously peptidomimetics (Vb), that act as a dopamine modulators, based on the same pyridine scaffold have been synthesized with the amino acid sequence Pro-Leu-Gly-NH2.64 The second amide bond isostere (X = O) in peptidomimetics

Vb lack the hydrogen bonding capability as compared to Va (X = NH) and

(33)

N H H N N H O O O R1 _N X OH O O H2N R1 IV Va X = NH R2 R3 R3 R2 Vb X = O

Figure 6.3 A peptidomimetic containing the pyridine scaffold corresponds well to a

peptide in an extended conformation, both according to electrostatic calculations and geometric fit.

Electrostatic calculations (Spartan software, AM1 Hamiltonian) on a peptidomimetic closely related to Va (R1 = R2 = R3 = Me, terminally N-acetylated and having a C-terminal NHMe) confirm a similar charge distribution for the amide bond isostere attached at position 2 of the pyridine ring (NH and pyridine N) and the corresponding amide bond in Ac-Ala-Ala-Ala-NHMe.64_{The side chain of amino acid 2 (corresponds to R}2_{attached at}

position 3 of the pyridine ring) in our mimetic was chosen to be a proton to avoid the risk for a steric clash between two R2 groups or between R2 and a peptide side chain in a β-strand aggregate (Figure 6.4). If desired, substituents at position 3 of the pyridine ring can be introduced by using highly reactive electrophiles in a halogen dance reaction.65

N N H O O N H R1 R2 R3 N H N O O H N R1 R2 R3 N N H O O N H R1 R2 R3 N H O H N O O H N R1 R2 R3

Figure 6.4 A hypothetical antiparallel β-sheet aggregation between two β-strand

mimetics Va (left) and between a peptide and β-strand mimetic Va (right).

Usually polar and nonpolar amino acid side chains alternate in native β-sheets. Therefore nonpolar side chains were selected for the R1- and R3 -groups in β-strand mimetic Va, intended to be iso-butyl (side chain of leucine) or sec-butyl (side chain of iso-leucine). A retrosynthetic analysis of

(34)

2-fluoro-4-N N H OH O O H2N R1 O PgHN R1 N H2N OH O H F I Va VI VII VIII H2N H N N H CO2H O O R1 IV R3 R3 R3

Figure 6.5 A retrosynthetic analysis of β-strand mimetic Va, based on substituted

pyridine VII as scaffold.

The aromatic substitution at position 4 of the pyridine ring was planned to take place via a Grignard exchange reaction using iso-propylmagnesium chloride to afford a pyridine Grignard reagent, using aldehyde VI as electrophile.66,67 Nucleophilic aromatic substitution (SNAr) at position 2 of

the pyridine ring, by displacing the fluorine atom using an amine as nucleophile, was planned as another key transformation. The preferred order to introduce the two substituents was to perform the Grignard exchange reaction first due to expected selectivity problems between the two halogenated positions of the pyridine ring in the SNAr reaction. In addition,

interference from an ‘anilinic’ proton in a strongly basic Grignard reaction was anticipated to cause problems.

6.3 Attachment of an N-terminal leucine analogue at

position 4 of the pyridine ring

It was reasoned that the Grignard reaction to connect protected leucinal at position 4 of the pyridine ring would suffer from interference from the acidic NH proton in the carbamate functionality at the α-amino group. However, an improved yield was expected if di-Boc-protected leucinal was used as electrophile (Scheme 6.1). As the first step in an attempt to prepare this compound, Boc-protected leucine was reduced to the corresponding alcohol

41 (97%) by activating the carboxylic acid with iso-butyl chloroformate

(35)

protection of the alcohol moiety in 41 as a silyl ether using TBDMSCl and imidazole in CH2Cl2 to afford 42 (90%). OH O NHBoc OH NHBoc i-Bu Chloroformate NaBH4 97% 41 OTBDMS NHBoc 42 TBDMSCl Imidazole CH2Cl2 90% 66% OTBDMS NBoc2 Boc2O DMAP dioxane 43 TBAF THF OH NBoc2 44 Oxidation O NBoc2 45

Scheme 6.1 The strategy to di-Boc-protected leucinal failed, probably during the

deprotection of 43.

In order to add the second Boc-group a large excess Boc2O and DMAP was

added to 42. The reaction was heated to reflux in dioxane which afforded the di-Boc-protected 43 (66%). Thereafter cleavage of the silyl ether in 43 by using TBAF was attempted. The 1_{H NMR spectrum confirmed removal of}

the silyl ether, but the signal from the Boc-group was now split into two, probably due to migration of one of the two Boc-groups to the primary alcohol. Because of some uncertainty regarding the compound identity, oxidation to the corresponding aldehyde 45 was tried using TEMPO, NaOCl and KBr in CH2Cl2.68 No traces of aldehyde 45 was detected, supporting that

a Boc-protected alcohol, which is unable to undergo oxidation, had been formed during the treatment with TBAF.

OH NHBoc OH NH2 OH N3 Oxidation O N3 80% Formic acid Tf2O NaN3 64% 41 46 47 48

Scheme 6.2 The synthetic strategy toward azidoaldehyde 48 failed in the oxidation

of azidoalcohol 47.

A second strategy aiming to avoid an acidic carbamate proton in the leucinal to be used in the Grignard reaction, involved the use of an azide as protection for the α-amino group. The synthetic route to azidoleucinal 48

(36)

observed. The neighboring azide might be responsible for the failure of these oxidation reactions. These disappointing results lead us to examine if the Grignard reaction could be performed in a reasonable yield in presence of the carbamate NH (Scheme 6.3). Dess-Martin periodinane oxidation of alcohol 41 generated the desired aldehyde 49 (88%). Due to the reported sensitivity of the chiral center in protected α-amino aldehydes, aldehyde 49 was used without further purification in the next step.72,73_{The following}

Grignard reaction was performed by adding iso-propylmagnesium chloride to 2-fluoro-4-iodopyridine thereby preforming the pyridine Grignard reagent.64,67_{Then aldehyde 49, dissolved in THF, was added to afford}

alcohol 50. Considering the acidic carbamate NH a two fold excess of both 2-fluoro-4-iodopyridine and iso-propylmagnesium chloride was used.

O NHBoc 49 iPrMgCl THF 2F,4I pyridine 41%, from 49 N F BocHN OH 50 51 N F BocHN OBn BnBr QHSO4 50% NaOHaq OH NHBoc 41 88% Dess-Martin periodinane

Scheme 6.3 A successful strategy to connect leucinal at position 4 in

2-fluoro-4-iodopyridine ring via a Grignard exchange reaction.

To our delight the secondary alcohol 50 was formed in a reasonable yield from this reaction as indicated by NMR spectroscopy of the crude material. Crude 50 was then protected as a benzyl ether under phase transfer conditions to afford 51.74 Fortunately, purification of benzyl ether 51 was straightforward (in comparison to alcohol 50) and the desired product was obtained in 41% yield (based on 49) over the two steps. The N-terminal leucine derivative was now properly connected to the pyridine scaffold, and the conjugate was protected in the desired manner.

6.4 Attachment of a C-terminal glycine analogue at

position 2 of the pyridine ring

6.4.1 Nucleophilic aromatic substitution

A model study to displace the fluorine in 2-fluoropyridine with an amine was performed using benzylamine (1.2 equiv.) as nucleophile. A variety of conditions were applied to establish a robust and high yielding SNAr

(37)

Table 6.1 Different reactions performed with benzylamine (1.2 equiv.) as

nucleophile in an SNAr reaction using 2-fluoropyridine as electrophile.

N F H2N + N N H 52

Entry Solvent Base/additive Temp. ° C

Time Heating method Yield % 1 NMP TEA 150 48 h Sealed cylinder 30 2 NMP TEA 250 30 min Microwave 42 3 NMP DIPEA 250 60 min Microwave 52 4 Pyridine 250 60 min Microwave 79 5 CH3CN Ag2CO3 120 10 min Microwave

6 CH3CN CuCO3 120 10 min Microwave

7 Toluene DIPEA/NaCl 200 20 min Microwave

8 THF n-BuLi 78→rt 3 h - -

9 THF n-BuLi/Ag Zeolite 78→rt 3 h

As revealed by entries 8 and 9 in table 6.1 strongly basic conditions based on

n-BuLi failed to generate product. In contrast, use of amine bases and

elevated temperatures, entries 1−4, afforded the target compound 52 in modest to excellent yields. The idea behind entry 5 was to get a coordination between Ag+_{and the fluorine attached to the pyridine ring, thereby}

attempting to enhance the reactivity of the electrophile. In entry 6 the possibility to have a coordination between Cu2+_{and the pyridine nitrogen}

was tested to get an increased reactivity. Toluene is not a good solvent in absorbing microwaves but adding solid NaCl generate hotspots around the salt where the reaction is thought to take place. In entry 7 this method was applied but no traces of product was observed.

As the next step of the exploration of the SNAr reaction the amine

nucleophile was chosen to be the sterically demanding Ile. We reasoned that if a method to substitute 2-fluoropyridine using a derivative of Ile could be developed, the method should also be applicable to several less sterically hindered amino acids.

iBu chloroformate

NaBH4 Formic acid

BnBr QHSO4 NaOHaq PhCH3

(38)

To prepare a suitable Ile derivative Boc-Ile-OH was reduced to the corresponding alcohol by first activating the acid with iso-butyl chloroformate followed by addition of NaBH4 to afford 53 (Scheme 6.4).55

Treating 53 with BnBr under phase transfer conditions, using QHSO4 as

phase transfer catalyst, afforded ether 54 (85%, from Boc-Ile-OH).74 Finally, the Boc-group was removed using formic acid to afford 55 (91%), which was used as nucleophile in the aromatic substitution of 2-fluoropyridine (Table 6.2).

Table 6.2 Different reactions performed with 55 (1.5 equivalents) as nucleophile in

the SNAr reaction using 2-fluoropyridine as electrophile.

N N H OBn N H2N OBn F + 55 56

Entry Solvent Base/additive Temp. ° C Time Heating

method Yield % 1 HMDS TEA/TMSCl 175 1 h Microwave

2 Pyridine 190 30 min Microwave 3 Pyridine 230 1 h Microwave 7 4 NMP pyridine 250 30 min Microwave

5 THF BuLi 78→ 0 3 h 40

6 THF MeMgCl Reflux 7 h Oil bath 38 7 THF NaHMDS/TMSCl 78→rt 5 h

8 THF

NaHMDS/15-crown-5 78→reflux 7 h Oil bath *

* Compound 56 was not formed, instead a nucleophilic attack from 15-crown-5 on 2-fluoropyridine occurred and gave compound 57 (Scheme 6.5).

Using 55 as nucleophile in this reaction generally afforded no product or low yields. In contrast to the results with benzylamine only strongly basic methods, entries 5 and 6, generated the target compound 56, but only in approximately 40% yields. Interestingly, treating 2-fluoropyridine with NaHMDS, 55 and 15-crown-5 ether in refluxing THF afforded a substitution product where the crown ether had displaced the fluorine atom to afford 57 (68%, Scheme 6.5). N F 15-crown-5 NaHMDS THF reflux 68% O O O O O N 57 + 55 H2N OBn

Scheme 6.5 Unexpected byproduct formed in an attempt to react 2-fluoropyridine

(39)

The conditions reported in entry 6, which gave a more stable reaction outcome than those of entry 5, were now applied in an SNAr reaction of

fluoropyridine 51 (Scheme 6.6). H2N OBn 55 51 N F BocHN OBn 58 N N H BocHN OBn OBn MeMgCl THF +

Scheme 6.6 A synthetic attempt to afford substituted pyridine 58 using the preferred

method obtained in the model study.

Unfortunately, the reaction towards 58 resulted in multiple undesired products and an alternative synthetic route was needed. First an investigation of if less sterically hindered amine nucleophiles could accomplish this transformation on substituted pyridine 51 was initiated. Different glycine analogues were used as nucleophiles and 51 as electrophile (Scheme 6.7). First, an excess of H2N-Gly-OtBu (5 equiv.) was used as nucleophile and

reacted with 51 at 150 °C in pyridine by using microwave irradiation. The reaction mixture turned brown but the desired product was only observed in trace amounts (LCMS). It was also found that the Boc-protection group was labile and partially cleaved under these conditions. Boc-protected 51 was therefore converted to the more stable acetamide 60 (86% from 51, Scheme 6.7).

(40)

N F + + + H2N OH H2N OtBu O H2N OH O 10 equiv. pyridine 190 !_C 1 h N N H OH N N H Allylamine 17 bar 2.5 h 5 equiv. 5 equiv. pyridine AcHN OBn 60 N F OBn AcHN N F OBn H2N N F OBn BocHN Formic acid Ac2O CH2Cl2 59 51 60 (86%, from 51) N N H OBn AcHN OtBu O 61, trace amounts NaHCO3 N F AcHN OBn 60 N N H OBn AcHN OH O + NH O OBn AcHN 62, < 10% 63, < 10% N F AcHN OBn 60 N F AcHN OBn 60 AcHN OBn AcHN OBn 64, 54% 65, > 90% crude A. B. C. D. E.

Scheme 6.7 The Boc-protection group was partially cleaved during attempted SNAr

reaction of 51 and was therefore converted to acetamide 60 (A). Thereafter, different glycine analogues were used as nucleophiles in the SNAr reaction with 60 (B→E).

An excess of H2N-Gly-OtBu (5 equiv.) and 60 were heated to 150 °C in

pyridine by using microwave irradiation. However, also in this case the reaction turned brown and 61 was detected only in trace amounts (LCMS). Raising the temperature to 180 °C did not increase the yield of 61, but resulted in formation of a black solid. The side reaction was believed to be due to polymerization of tert-butyl protected glycine. Actually, when the reaction was run without fluoropyridine 60 a similar black solid was formed as well. As the problem thus seemed to originate from the tert-butyl ester, glycine with the carboxylic acid moiety unprotected was instead tested as nucleophile. Solubility problems of glycine in organic solvents required the use of aqueous media in the reaction and saturated aqueous NaHCO3 was

chosen. Subsequently, excess glycine (5 equiv.) was heated with 60 to 160 °C for 1 h using microwave irradiation. Indeed the desired product 62 was formed but in small amounts (< 10%) as revealed by LCMS analysis. Unfortunately, a byproduct resulting from an attack of water to afford

(41)

pyridone 63 (LCMS) was also formed in equimolar amounts. Our interest therefore turned to other glycine analogues with better solubility in organic solvents in order to eliminate water as competing nucleophile. Ethanolamine and allylamine were chosen since both contain functionalities that could serve as masked carboxylic acids. First excess ethanolamine (10 equiv.) was heated together with 60 in pyridine to 190 °C for 1 h. This gave the desired substitution product 64 in 54% yield. When evaluating different experimental conditions it was noticed that the yields of the nucleophilic aromatic substitutions was increased by three factors. High concentrations of the amine nucleophile and long reaction times at high temperatures. Therefore, allylamine was used as solvent in the SNAr reaction with 60, and

the reaction was run at 17 bar (~150 °C) for 2.5 h using microwave irradiation. To our satisfaction, all of 60 was cleanly converted to the desired product 65, which was obtained in high yields (> 90%).

With the substitution using ethanolamine and allylamine as nucleophiles accomplished in good to excellent yields the next task was to oxidize the alcohol moiety in 64 or the olefin in 65 to the desired carboxylic acid, or preferably to the corresponding ester to avoid zwitterionic compounds. The development of oxidation conditions was not performed on substituted pyridines 64 and 65, but instead on the more simple compounds 66 and 67 generated from 2-fluoropyridine in a related manner as described above (Scheme 6.8, for details see paper III).

N N N H N H OH N N H OH O N N H RuCl3 NaIO4 OH OH K-osmate NMO 51% 66 67 N N H N N H 67 O3 NaOH MeOH OMe O N N H OMe O NaIO4 Br2 MeOH 70 70 69 68

Scheme 6.8 Attempted oxidation to give the desired carboxylic acid 68, or

(42)

acid 68, assumed to be a zwitterion, were not explored further due to anticipated problems in handling the charged target compound. Methods to directly generate ester 70, thereby avoiding the carboxylic acid, seemed more appealing. First, oxidation of the olefin in 67 to the corresponding diol

69 (51%) was accomplished by using a catalytic amount of potassium

osmate with NMO as cooxidant. The diol was further treated with Br2 and

NaIO4 in methanol, but the desired product was not observed.76 A smooth

procedure for oxidation of olefins directly to the corresponding methyl ester has been reported. This method was applied to olefin 67 by purging a stream of ozone through a solution of 2 M methanolic NaOH and CH2Cl2 containing

67.77 Unfortunately, in our hands all 67 was consumed but no product was observed by TLC or NMR analysis of the crude product.

6.4.2 A reductive amination strategy

Treating 2-aminopyridine with a strong base (e.g. NaH) followed by alkylation is known to alkylate the nitrogen in the pyridine ring instead of the amine in position 2.78 On the other hand applying reductive amination conditions is known to result in alkylation of the amino group.79_{Therefore a}

model study was performed using 2-aminopyridine and anisaldehyde under various reductive amination conditions (Scheme 6.9). Acidic conditions using AcOH and Na(OAc)3BH in CH2Cl2 or MeOH, neutral methods in

methanol employing molecular sieves as drying agent and a basic method using Na(OAc)3BH and TEA in CH2Cl2 were investigated. The basic

conditions49_{turned out to be superior but still afforded 71 only in a modest}

36% yield. To afford the desired glycine substituted pyridine 68 the conditions from the model study were applied to 2-aminopyridine and glyoxylic acid, but only starting material was recovered from the reaction. The more soluble tert-butyl glyoxylic acid, generated from tert-butyl acrylate via ozonolysis, was also employed in the reductive amination. 1H NMR spectroscopy of the crude product revealed that, most likely, the tert-butyl ester of α-hydroxy acetic acid was formed as the major product.

N NH2 N N H OMe N NH2 N N H OH O or N N H OtBu O Na(OAc)3BH TEA DCE Reductive amination 68 71 36% 72

Scheme 6.9 Attempts to afford glycine substituted pyridine via reductive amination

(43)

Another approach to afford the glycine substituted pyridine 68 by refluxing 2-aminopyridine with concentrated perchloric acid and glyoxal in methanol has been described in the literature.78_{In an attempt to use this approach}

aminopyridine 73 was prepared in 60% yield by heating 60 (~150 °C, sealed steel cylinder) in water containing 25% ammonia (Scheme 6.10).80 Then 73, perchloric acid and glyoxal were heated to reflux in methanol, but unfortunately these rather harsh conditions resulted in decomposition of the starting material without any detectable formation of 74.

N NH2 OBn AcHN N N H OBn AcHN OMe O perchloric acid glyoxal 25% NH3 in H2O 60% 73 74 N F OBn AcHN 60

Scheme 6.10 An SNAr reaction displacing the fluorine with NH3 followed by an

attempt to afford glycine substituted pyridine 74.

6.4.3 Changing the substitution order and starting with the S

N

Ar

reaction

As previously mentioned the preferred order of substitution of the pyridine scaffold was to perform the Grignard reaction prior to the SNAr reaction.

This was due both to a possible selectivity issue between the two halogenated positions in 2-fluoro-4-iodopyridine and an anticipated interference from the ‘anilinic’ proton when running the Grignard reaction as the second step. If both selectivity and protection of the ‘anilinic’ proton prior to the Grignard reaction could be accomplished, reversal of the reaction order could be successful. This approach was investigated as outlined in Scheme 6.11.

(44)

N I F N I N H heat allylamine TsCl + H2N N H Ts 2.5 equiv. CH2Cl2 99% N I N Ts 63% DMSO KH heat 2-F,4-I-pyridine iPrMgCl 49 N N Ts OH BocHN 77 76 78 75

Scheme 6.11 Selective nucleophilic aromatic substitution of 2-fluoro-4-iodopyridine

using a deprotonated sulphone amide as nucleophile.

To begin with, 2-fluoro-4-iodopyridine and allylamine were heated (180−200 °C) using microwave irradiation, but the mono substituted product

75 was not observed. This may, at least partly, be due to selectivity

problems. Allylamine was therefore transformed to tosyl amide 76 to get a more acidic NH proton, thereby facilitating the formation of a nitrogen anion. When 76 and 2-fluoro-4-iodopyridine were treated with KH in DMSO (140 °C, 30 minutes) using microwave irradiation, iodopyridine 77 was obtained in acceptable yield (63%) with the ‘anilinic’ nitrogen atom protected. When using other solvents (THF, DMF or H2O) or other bases

(Cs2CO3 or NaH) no product was formed, suggesting that the anion of

DMSO is the best choice of base to be employed in this type of SNAr

reaction. Continuing with the Grignard exchange reaction using iso-propyl magnesium chloride at room temperature followed by addition of aldehyde

49 failed to give the desired product 78. Instead, the deiodinated starting

material was observed as the major product in the reaction mixture (LCMS). Thus, it seemed like the preferred order to perform the substitutions on the pyridine scaffold still was to start with the Grignard reaction.

6.5 Completing the synthesis − A successful Boc

strategy

The ‘anilinic’ proton, present in several of the synthetic approaches described above, was believed to be responsible for the failures in completing the synthesis so far. Protection of the ‘anilinic’ nitrogen as a tosyl amide described in the preceding section lead us to the idea to use a Boc-protection strategy in the last steps of the synthetic sequence. Therefore

(45)

allylamine substituted pyridine 67 was treated with Boc2O and a catalytic

amount of DMAP in CH2Cl2 which allowed 79 to be isolated in excellent

yield (99%) (Scheme 6.12).79 N N H N N Boc Boc2O DMAP 99% 67 79 65% O3 NaOH MeOH N N Boc 80 OMe O

Scheme 6.12 Protecting the ‘anilinic nitrogen allowed oxidation of the alkene.

To our delight, purging ozone through a CH2Cl2 solution containing 2 M

methanolic NaOH and 79, now afforded 80 in acceptable yield (65%).77_As

described previously oxidation of 67 in the same manner failed to give the desired ester.

With all parts in hand, the synthesis of the β-strand mimetic could now be brought to completion. This started with the SNAr reaction, which was

performed by heating 60 to 17 bar (∼150 °C) using microwave irradiation in neat allylamine (Scheme 6.13). Excess allylamine was then removed and amine 65 was Boc-protected using excess Boc2O and a catalytic amount of

DMAP in CH2Cl2 to afford 81 (86%, from 60).79 The olefin was treated with

ozone in a CH2Cl2 solution containing methanolic NaOH to afford ester 82

(58%).77_{Here an unwanted oxidation of the benzyl ether to a benzoyl ester}

was observed, but adjusting the reaction time carefully minimized this side reaction. Next, the benzyl ether was subjected to hydrogenation to afford alcohol 83 (75%). To eliminate the risk for epimerization in the following step, during oxidation of the alcohol, Dess-Martin periodinane was employed to afford ketone 84 (81%).71_{Removal of the Boc-group was}

thereafter accomplished in neat formic acid to afford β-strand mimetic 85 (89%). Surprisingly, when 85 was analyzed by chiral chromatography it was found that a partial epimerization of the stereogenic center (60% ee). Fortunately, chiral chromatography of Boc-protected 84, the direct precursor of β-strand mimetic 85, showed no detectable epimerization which proved that the synthetic steps used to obtain 84 had not caused the epimerization. Evidently, the epimerization of 85 had occurred during the acidic conditions applied in the last step to remove the Boc-group.

(46)

N F AcHN OBn N N H AcHN OBn N N AcHN OBn Boc N N AcHN OBn Boc OMe O N N AcHN OH Boc OMe O N N AcHN O Boc OMe O N N H AcHN O OMe O

Allylamine BocDMAP2O

86%, from 60 O3 NaOH MeOH 58% H2 Pd/C MeOH 75% Dess-Martin periodinane 65 81 82 83 84 85 60 Formic acid 81% 89%

Scheme 6.13 Synthesis of the β-strand mimetic 85. Chiral chromatography of 85

revealed that partial epimerization had occurred in the last step.

Our first idea to obtain 85 as a single enantiomer was to reverse the order of the two last synthetic steps. Application of acidic conditions to alcohol 83, instead of to the enolizable ketone 84, would eliminate the risk for epimerization (Scheme 6.14). Consequently, alcohol 83 was treated with TFA (25%) in CH2Cl2 to liberate the amine affording 86, directly followed

by oxidation using Dess-Martin periodinane for 3 minutes with a reductive workup. This gave β-strand mimetic 85 (66%, from 83) without any detectable epimerization. In this way the synthesis of β-strand mimetic 85 was completed over a twelve step synthetic sequence with an overall yield of 7%. N N Boc OMe O AcHN OH N N H OMe O AcHN OH N N H OMe O AcHN O TFA CH2Cl2 Dess-Martin periodinane 83 86 85 66%, from 83

Scheme 6.14 By reversing the order of the two final steps, 85 was afforded as a

Synthesis of β-turn and pyridine based peptidomimetics