MSc Chemistry
Molecular sciences
Master literature thesis
Macrocyclization of unprotected peptides and
proteins using native amino acids
by
Marnix George Roseboom
VU ID: 2574728
UvA ID: 12033839
June 2020
12 credits
may 2019-june 2020
Supervisor/Examiner:
Examiner:
Supervisor: Prof. dr. T. N. Grossmann
Dr. S. Hennig
3
Abstract
Peptides and proteins are a class of biomolecules that are gaining more attention due to their wide scope of applications such as enzymes, bioimaging, and therapeutics. However, applying peptides for these technologies is a difficult task, as numerous problems such as instability, low solubility, low cell permeability or insufficient folding occur. To circumvent these complications, various strategies have been applied. A notable method is by utilizing peptide macrocyclization, which includes the linking of residues or chain end with side chains to form a macrocycle. There are roughly three types of macrocyclization: by applying Tags, utilizing non-proteinogenic amino acids or by installing biselectrophilic moieties between native amino acids. The latter displays advantages over the other two methods, as this approach allows design of the linker’s nature to finetune features such as solubility and stability. Additionally, the linker can be designed to tolerate follow up chemistry to further modify the macrocyclic peptide. Optimization of the macrocyclization reaction may tolerate peptide constraining without the use of protecting groups on the residues.
This thesis summarizes the main classes of cross-linkers that are applied in macrocyclization of peptides and proteins, including halobenzyls, haloalkyls, conjugate acceptors, maleimides, haloaryls and aryl palladium reagents.
4
List of abbreviations
ACE2 Angiotensin-converting enzyme 2 ACM acetamide
Aib α-aminoisobutyric acid
Bph 4,4′-bis(bromomethyl)biphenyl Bpy 6,6’-bis-bromomethyl-[3,3’]bipyridine
BSBCA 3,3'-bis(sulfonato)-4,4'-bis(chloroacetamido)azobenzene
BSBDA 3,3′-bis-(Sulfonato)-4,4′-bis(buta-2,3-dienoylamido)azobenzene) CD circular dichroism
CovCore covalent linked core
Cα Alpha carbon/central carbon
DA Diels-Alder
DBMB bis(bromomethyl)benzene DCA dichloroacetone
DCC N,N'-dicyclohexylcarbodiimide DEAD diethyl azodicarboxylate DIPEA N, N-diisopropylethylamine DMSO dimethyl sulfoxide
DTT dithiothreitol
DVT divinyltriazines Et3N triethylamine
FDA Food and Drug Administration Fmoc fluorenylmethyloxycarbonyl GSH glutathione
5
HPLC high-performance liquid chromatography
In situ on site
In vitro in glass
In vivo in life
K2CO3 potassium carbonate
MeCN acetonitrile
MPAA mercaptophenylacetic acid MS mass spectrometry
NBD nitrobenzofurazan NCL native chain ligation
NMR nuclear magnetic resonance PDB Protein Data Bank
PDI peptide dual inhibitor PEG poly(ethylene glycol)
PET allowing Positron Emission Tomography PPh3 triphenyl phosphine
RCM ring-closing metathesis
SnAr nucleophilic aromatic substitution
SPECT Single Photon Emission Computed Tomography SPPS Solid-phase peptide synthesis
TBMB tris(bromomethyl)benzene TCEP tris-(2-carboxyethyl)phosphine TERMS tertiary motif package
TFA trifluoroacetic acid
Tris tris(hydroxymethyl)aminomethane
6
Table of content
1 Introduction ... 7
2 Halobenzyls as cysteine cross-linkers ... 11
2.1 Mono halobenzyl cross-linking on cysteines ... 11
2.2 Extended halodiaryls targeting cysteines ... 12
2.3 Trivalent benzyl linkers ... 14
3 Haloalkyl substitutions ... 18
3.1 Methylene linkers on cysteines ... 18
3.2 Oxetanes as cysteine linkers ... 19
3.3 Acetone-based cysteine linkers ... 21
3.4 Chloroacetamide linkers for cysteine-tail macrocyclization ... 23
3.5 Iodoalkyl-based photoswitch linkers targeting cysteines ... 25
4 Conjugated cross-linkers and maleimides ... 26
4.1 Cysteine connected allene-based photoswitch linkers ... 26
4.2 Meldrum’s acid derived linkers for cysteine-N-term linkage ... 27
4.3 Macrocyclization and functionalization using divinyltriazines on cysteines ... 29
4.4 Bromo- and thiophenol maleimide linkage of cysteines ... 32
4.5 Dibromomaleimide insertion into disulfides and post cyclization functionalization ... 32
5 Haloaryls as cross-linkers by nucleophilic aromatic substitution ... 37
5.1 Perfluoroaromatic monoaryls and bisaryls as disulfide linkers ... 37
5.2 Lysine-targeted N-arylation with haloaryls ... 39
5.3 Extended perfluoroaromatic aryls targeting cysteines... 41
5.4 S-tetrazines as reversible cysteine linkers ... 43
6 Palladium mediated cross-linking of aryls on residues ... 46
6.1 Palladium mediated arylation of cysteines... 46
6.2 Palladium mediated arylation of lysines ... 47
7 Concluding words ... 49
7
1 Introduction
Amino acids are small organic molecules that contain an amine (-NH2), a carboxylic group (-CO2H) and
a central carbon (Cα).1 The feature that distinguishes each amino acid from another is the variation in
side chains attached to the Cα. A total of 20 amino acids are encoded by the human genetic code and
some uncommon additional amino acids occur due to enzymatic post-translational modifications.1
Amino acids are also called the “building blocks of life”, as many scientists believe that the origin of life on earth had started with the synthesis of these building blocks.2 Nearly all the proteins and peptides
in the human body are made from the 20 encoded amino acids via end-to-end connections called peptide bonds (Figure 1). Multiple connections of amino acids elongate the chain to make a polypeptide, of which the long chain of peptides is called the “backbone”.1
Figure 1: General structure of amino acids in a polypeptide chain.
The variety of amino acid residues, along with the rotational bonds of C-N(amine) and C-C(carboxyl) allow the folding of the linear polypeptide chain into defined three-dimensional structures.1
(Bio)chemists and biologists distinguish protein structures on four different levels: primary, secondary, tertiary and quaternary (Figure 2)1 The primary structure solely focuses on the amino acid sequence,
while the secondary structure refers to the three-dimensional motifs of a local polypeptide sequence. Commonly found secondary structures are α-helices and β-sheets, which are formed by intramolecular (non-)covalent interactions (Figure 2). Tertiary protein structures include the overall three-dimensional structure of a protein and the quaternary structure is an assembly of multiple folded polypeptides.1 An
example of the 4 structure levels of the 1AXC protein assembly is displayed in Figure 2 (obtained from PDB).3
Figure 2: The primary, secondary, tertiary and quaternary structures of the 1AXC protein (images obtained and modified
8
Proteins undergo protein-protein interactions (PPIs) or interactions with smaller peptides and small molecules, triggering molecular transformations or cascade responses.4 For example, the cytochrome
P450 enzyme superfamily act as monooxygenases on various small molecules such as drugs, increasing hydrophilicity and promoting actions like excretion.5 A second example includes the interaction of the
somatostatin receptor with its endogenous ligand somatostatin. This protein-peptide interaction triggers a cascade of additional protein interactions, regulating the endocrine system and processes such as cell proliferation.6 Lastly, the PPI of human angiotensin-converting enzyme 2 (ACE2) with
surface glycoproteins (peplomers) of the SARS-CoV-2 virus allows entering of the host cell, leading to infection and disease development.7
These wide scope of protein characteristics have inspired (bio)chemists to design, synthesize and apply proteins and peptides for specific use. Lately, developments in protein engineering have allowed rational and irrational design of enzymes to perform highly chemoselective reactions. In 2018, chemical engineer Francis Arnold was awarded with the Nobel Prize in chemistry for development of directed evolution as strategy to obtain novel enzymes.8 Another field of protein engineering has led
to the design and synthesis of antibody-based tracers. Here, antibodies are provided with a linker attached to a radioactive nuclide, allowing Positron Emission Tomography/Single Photon Emission Computed Tomography (PET/SPECT) imaging.9 These techniques are employed to obtain important
information about a certain disease, such as locating tumours in cancer patients.
Recently, developments have been made in the research field of peptide-based medicine. In the past, peptides were considered ‘undruggable’ due to their low capability of cell-entering and low serum stability.10 However, a recent trend has shown that peptide-based drugs are now emerging, as novel
peptide engineering methods are being employed to circumvent these problems.10 Table 1 displays
the increasing number of Food and Drug Administration (FDA) approved therapeutic peptides from years 2000-2014.10
Table 1: FDA approved peptide-based drugs from 2000-2014.10
Name Year of approval Therapeutic area Name Year of approval Therapeutic area
Atosiban 2000 Obstetrics Mifamurtide 2009 Oncology
Taltirelin 2000 CNS Liraglutide 2009 Metabolic disease
Aviptadil 2000 Urology Tesamorelin 2010 Antiinfective
Carbetocin 2001 Obstetrics Lucinactant 2012 Pulmonary Nesiritide 2001 Cardiovascular Peginesatide 2012 Hematology Teriparatide 2002 Osteoporosis Pasireotide 2012 Endocrinology Enfuvirtide 2003 Antiinfective Carfilzomib 2012 Oncology
Abarelix 2003 Oncology Linaclotide 2012 Gastroenterology
Ziconotide 2004 Pain Teduglutide 2012 Gastroenterology
Pramlintide 2005 Metabolic disease Lixisenatide 2013 Metabolic disease Exenatide 2005 Metabolic disease Albiglutide 2014 Metabolic disease Icatibant 2008 Hematology Oritavancin 2014 Antiinfective Romiplostim 2008 Hematology Dulagutide 2014 Metabolic disease Degarelix 2008 Oncology Afamelanotide 2014 Dermatology
9
The discovery of solid-phase peptide synthesis (SPPS) by Merrifield has been a breakthrough in the synthesis and design of peptides.11 This method allows synthetic (bio-)chemists to synthesize peptides
of interest in a straightforward approach. A typical peptide synthesis is performed with fluorenylmethyloxycarbonyl (Fmoc) protected amino acids (Scheme 1). Synthesis starts with the linking of the first amino acid with a polymeric resin bead.12 Carboxyl activation is done with
N,N'-dicyclohexylcarbodiimide (DCC). Once the amino acid is connected to the bead, Fmoc deprotection is accomplished with bases (usually piperidine). A second amino acid is added and activated with DCC in presence of hydroxybenzotriazole (HOBt). HOBT prevents isomerization of the peptide backbone.13
The process of Fmoc deprotection and amino acid linking is repeated until the desired sequence is synthesized. Final cleavage of the peptide and bead is usually accomplished using trifluoroacetic acid (TFA).
Scheme 1: A general strategy of Fmoc-based SPPS.
SPPS has provided facile access to peptides for further research towards potential new medicines. However, many studies aiming to explore bioactivity require folding of proteins and peptides, as function is linked to its three-dimensional structure.1 In ribosomal polypeptide synthesis, folding is
assisted by a group of large proteins called chaperones.14 Linear peptides do not have a defined
structure and presence of chaperones help to overcome the entropic penalty of the folding process. When applying SPPS, peptide folding is not taking place naturally as chaperone proteins are lacking. Lately, multiple strategies have been developed to control folding of synthetically obtained peptides. One of these strategies include macrocyclization, which provides a covalent connection between residues that are distant in primary sequence but close in native three-dimensional structure.15
10
Macrocyclization can be achieved in roughly three methods: protein/enzyme-mediated cyclization, incorporation of non-native amino acids and insertion of dielectrophilic linkers on native amino acids (Figure 3).15, 16, 17 One example of protein-mediated macrocyclization is the SpyTag strategy. This
method involves conjugation of recombinant proteins, where a SpyTag peptide is spontaneously connected to the SpyCatcher protein.16 The SpyTag and SpyCatcher sequences can be introduced in
DNA sequences in the protein of interest to form a fusion protein. Macrocyclization by utilizing non-proteinogenic amino acids has proven to be a versatile strategy.15 This is due to the large possible
residue designs on which numerous macrocyclization reactions can take place. Some extensive explored methods include a ‘click’-reaction between an azide- and alkyne amino acid, or a ring-closing metathesis (RCM) between two olefin-containing residues.17, 18 Macrocyclization by insertion of a linker
between residues often involves a double nucleophilic attack of two cysteine thiols on a biselectrophilic moiety.19 These cysteines often originate from naturally occurring disulfides. However,
macrocyclization on other residues or the tail/chain of the peptide are also known.
Figure 3: Macrocyclization via SpyTag approach, ‘click’-chemistry and RCM with non-proteinogenic amino acids and
insertion of biselectrophiles on native amino acids.
An advantage of the last mentioned method is that the nature of the linker can be modified to optimize physical chemical features of the protein/peptide such as solubility, cell permeability or stability.19
Another advantage is the fact that once the cross-linking reaction is optimized, macrocyclization can usually be performed on unprotected peptides. In contrast, some challenges of this strategy include tuning of the chemoselectivity towards the desired residues. If two cysteines are targeted, it must be certain that other nucleophilic side chains such as lysine or serine do not interfere with the cyclization. Additionally, the relevant reactive residues in proteins may be sterically hindered, complicating the cross-linking.20
The following chapters will focus on numerous categories of cross-linkers targeting specific residues. Here, the strategies will be illustrated in more depth, highlighting featured problems and how they are addressed, starting with halobenzyl cross-linkers targeting cysteines. In each example, the peptide has been synthesized by SPPS unless stated otherwise.
11
2 Halobenzyls as cysteine cross-linkers
2.1 Mono halobenzyl cross-linking on cysteines
In 2012, the group of Greenbaum reported the design and synthesis of a small constrained peptide, which was folded in its α-helix conformation by macrocyclization.21 The peptide was designed to
interfere with the PPI of calpain and its endogenous inhibitor calpastatin. Calpain belongs to the Ca2+
-dependent family of cysteine proteases and are involved in numerous pathological biological processes.22,23 At the time, no protease inhibitor had been reported with a constrained α-helical motif.
Peptide design initiated with analysis of the newly reported cocrystal structure of calpain with bound calpastatin.24 The crystal structure revealed that calpastatin is in its α-helical conformation, while in
solvent the protein does not have a defined structure. Further analysis shows that a small two-turn α-helix consisting of the sequence [IPPKYRELLA] is crucial for binding calpain. However, the [IPPKYRELLA] fragment itself is not capable of inhibiting calpain, as the helical content of this sequence in solvent is low. It was envisioned that the entropic penalty to form an α-helix from a random coil was too high, limiting the ability of protein inhibition.
Previous attempts for constraining the α-helix conformation included the use of non-native amino acids to perform reactions such as RCM,25 click-chemistry26 and lactam formation.27 Inclusion of
semi-rigid linkers showed potential and was successful in this method. The active site of calpains is highly conserved and insertion of non-native amino acids may form a challenge. With this information, the authors screened 24 different commercially available linkers on a model peptide (Ac-YGGEAAREACARECAARE-CONH2,1) containing 2 cysteines at the i, i+4 position. Of the 24 screened
linkers, the m-xylyl cross-linker provides peptide 2, yielding the highest helical content, determined by circular dichroism (CD) and nuclear magnetic resonance (NMR) spectroscopy (Scheme 2).
Scheme 2: Linker screening on the model peptide, leading to the discovery of the xylyl cross-linker to yield helical peptide 2.
With the identification of m-xylyl as an i, i+4 stabilizing cross-linker, further design of calpain was utilized. This time, the method was applied to the [IPPKYRELLA] sequence. Three [IPPKYRELLA] double cysteine mutants (3-6) were synthesized, along with the i, i+4 m-xylene cross-linked variants (4-6, Scheme 3).
12
Scheme 3: Synthesis and cross-linking of the [IPPKYRELLA] mutants.
Of the six peptides, the uncyclized variants showed little to no α-helicity, while stabilized peptides 7 and 8 showed improved helicity. Peptide 8 displayed by far the greatest α-helicity and kinetic studies showed that this peptide behaves as a competitive inhibitor of calpain. Continued modifications successfully lead to the synthesis from 8 to an activity-based probe, used to further explore the activity of calpains. These results demonstrate that monobenzyls can effectively be employed as α-helix stabilizer by macrocyclization.
2.2 Extended halodiaryls targeting cysteines
Besides monoaryls, bisbenzyls have also been utilized for macrocyclization of cysteine-containing peptides. Lin et al. described the use of 4,4′-bis(bromomethyl)biphenyl (Bph, 9) and
6,6’-bis-bromomethyl-[3,3’]bipyridine (Bpy, 10) as extended cross-linker (Figure 4).28, 29 As development for
this macrocyclization technique, the group performed a dynamic simulation study, revealing that the space between two cysteine sulphurs of a peptide dual inhibitor (PDI) at the i, i + 7 position in α-helix conformation ranged from 10.5 to 12.2 Å.29 Here, the cysteine carbon-carbon distance was 11.9 Å.
This conclusion lead to the screening of various aryl methylene bromides, of which Bph (9) and Bpy (10) provided a nearly perfect fit (Figure 4).
13
Figure 4: structures of Bph (9) and Bpy (10) cross-linkers (left) and modelling of Bpy (10) in α-helix conformation, adapted
and modified from Lin et al.29
This study revealed that the cysteine i, i + 7 macrocyclization with Bpy (10) and Bph (9) moderately enhances the α-helicity and significantly improves cell permeability without apparent cytotoxicity. These factors prove that the i, i + 7 bisbenzyl macrocyclization technique on cysteines have potential for peptide or protein delivery into cells.
Later, the group of Lin employed this macrocyclization technique as development for a stabilized peptide as potential cancer treatment.28 The peptide targets Mcl-1, which is a key protein in cancer
cell survival. BH3 proteins are pro-apoptotic factors and induce cell death by selectively binding to proteins like Mcl-1. The peptide NoxaB-(75−93)-C75A (Noxa peptide [AAQLRRIGDKVNLRQKLLN], 11) is derived from BH3 and had been modified by the earlier mentioned i, i + 7 disulfide macrocyclization method using Bph (9) as cross-linker (Scheme 4 A). Noxa peptide lacks two cysteines, thus it was decided to replace two solvent exposing residues with cysteines at i, i + 7 position followed by Bph cross-linking to yield macrocyclized peptide 13. Analogues were made by implementing D- and L-cysteines, leading to the [AA-(D)C-LRRIGD-(L)C-VNLRQKLLN] (12) cross-linked sequence. The group solved the crystal structure of the cross-linked sequence in mouse Mcl-1, where it was revealed that the cross-linked peptide 13adopts an α-helix conformation in the binding site (Scheme 4 B).
14
Scheme 4: A: cross-linking of peptide 12 with Bph 9 to obtain helical peptide 13. B: The crystal structure of bound peptide
13, confirming the helical structure, obtained and modified from Lin et al.28
Competitive fluorescence polarization (FP) assays demonstrated that the cross-linked sequence 13has a 12-fold increased inhibitory activity compared to the unstabilized Noxa peptide 12. Furthermore, the obtained crystal structure helped gaining insight of the binding mode, allowing further rational design of the cross-linked peptide. Eventually, peptide 13was modified by backbone methylations to create a higher helicity content, higher cell permeability, proteolytic stability and cell-killing activity.
So far, it has been demonstrated that mono- and bisbenzyls are suitable cross-linkers for i, i + 4 and i,
i + 7 positioned cysteine sulfides to enhance helicity and therefore improve therapeutic characteristics.
However, more recent works show that benzyls may also be implemented as a cysteine linker to obtain macrobicyclic peptides from its linear precursor.
2.3 Trivalent benzyl linkers
Heinis et al. reported the application of bis(bromomethyl)benzene (DBMB, 14) and tris(bromomethyl)benzene (TBMB, 15) for the formation of a bicyclic system.30 The research started
with the identification of sequences to selectively inhibit the urokinase-type plasminogen activator (uPA), which is a protein involved in tumour growth and invasion.31 Previously, it was displayed that
15
macrocyclic peptides exhibit inhibitory effects against uPA.32 These macrocyclic peptides show high
affinity and selectivity for uPA and it is believed that the peptide binds as one whole loop. However, the potency of these inhibitors is low and the peptides are therefore not suitable as therapeutics yet. To improve the potency of these linear precursors, it was envisioned that constraining the peptide single loop with TBMB (15) as trivalent linker would form an enhanced bicyclic peptide.
Figure 5: Structures of cross-linkers DBMB 14 and TBMB 15.
While screening a large library (4.0 × 109 peptides) of peptides, one hit displayed promising features.
This sequence (H-ACSRYEVDCRGRGSACG-NH2) was synthesized and macrocyclized with TBMB 15 to
create bicyclic peptide 16(Figure 6). As comparison, a monocyclic variant was synthesized where the second cysteine is exchanged with a serine and cyclized with DBMB 14 to yield peptide 17. Moreover, a third modified linear peptide 18 was synthesized where all cysteines were replaced with serines.
Figure 6: Bicyclic, monocyclic and linear peptide variants of the H-ACSRYEVDCRGRGSACG-NH2 sequence, obtained from
Heinis et al.30
Bioassays were performed where the three peptides were incubated with uPA, revealing that the degree of peptide constraint is proportional to the degree of uPA inhibiting potency. This means that the order of bioactivity drops with decreasing stabilization; bicyclic 16>monocyclic 17>linear18.
To investigate the binding mode of the bicyclic inhibitor of uPA, an X-ray crystal of the peptide-protein complex was solved (Figure 7). This revealed that a large area of the constrained peptide interacts with the uPA binding site, which preorganizes the peptide’s conformation to make the peptide-protein interaction more favourable. Compared to the most potent reported monocyclic uPA inhibitor, the novel bicyclic peptide displays a 200-fold higher degree of potency.
16
Figure 7: Crystal structure of bicyclic peptide 16 bound to uPa (UK18=novel bicyclic peptide), obtained from Heinis et al.30
To summarize, the utilization of these trimethylbenzenes is a successful method towards bismacrocyclization and preorganization of bioactive peptides. Furthermore, Implementation of trimethylbenzenes has also been performed for with higher peptide secondary structure complexity. In 2017, Degrado et al. published the de novo design and chemical synthesis of trimethylbenzene constrained non-natural peptide scaffolds for peptide-packing.33 Design of novel peptides often relies
on methods such as hydrophobic packing, metal binding or disulfides.34 However, this work
approaches protein-packing by implementing a covalent linked core (CovCore), which folds the peptide by a covalent link while simultaneously creating a hydrophobic core. The CovCore is used to constrain a de novo designed C3 symmetric peptide.
Design of the non-natural peptide was done by utilizing Rosetta together with the tertiary motif package (TERMS). Rosetta software is specialized in protein-protein docking and novel protein design.35 Earlier in this research, the group of Degrado was successful in design and synthesis of a C2
symmetric non-natural peptide consisting of two α-helices, cross-linked with ortho-dimethylbenzene at two cysteine residues.
Linear C3 precursor 19 was designed to include sequences coding for three α-helices containing a cysteine each (Scheme 5). First, precursor 19 was macrocyclized by backbone linking using 4-mercaptophenylacetic acid (MPAA) and dithiothreitol (DTT) to yield monocylic peptide 20, which would be followed by TBMB 15crosslinking on three cysteines to create tricyclic peptide 21. Synthesis of bicyclic peptide 22 was done with a similar strategy, except the backbone linking step had not been applied.
17
Scheme 5: Synthesis approaches towards peptides 20, 21 and 22 from linear precursor 19, adapted and modified from
Degrado et al.33
Analysis of the four synthesized peptides was done by mass spectrometry (MS), CD spectroscopy and NMR. CD spectroscopy analysis revealed that the helical content of the four peptides increased with the degree of constraining. This means that the linear precursor 19 contains the lowest helical content, followed by monocyclic 20, bicyclic 22 and tricyclic 21. The high helical content for peptide 21 is a validation of the performed Rosetta modelling. Further NMR analysis also reveals that tricyclic peptide
21shows full C3 symmetry, which is also in agreement with its designed structure. With the successful synthesis of 21, the researchers remodelled this peptide to create higher stability and less hydrophilicity in order to facilitate crystallization. Melting point analysis demonstrated that a higher degree of cyclization (tricyclic>bicyclic>monocyclic>linear) creates higher thermal stability.
To sum up, halobenzyls have shown great potential as cysteine side-chain linkers in order to preorganize peptides in α-helix conformation or peptide loops. Halobenzyls are useful stabilizers for rational peptide design, either as extended i, i+7 linkers or to synthesize bi- or tricyclic structures. A second class of cross-linkers are haloalkyls. In contrast to halobenzyls, haloalkyls often display other characteristics than enhancing the helical secondary structure. Furthermore, haloalkyls have shown potential as peptide cross-linkers with follow up functionalities, which is discussed in more depth in the next section.
18
3 Haloalkyl substitutions
3.1 Methylene linkers on cysteines
A common problem for disulfide-containing therapeutic peptides is the disulfide instability in reductive environments with for example hydrogenases, thiol oxidoreductases and disulfide isomerases.36 A
strategy has been applied where methylene is inserted into disulfides to increase the stability of peptides in these environments.37
In 2016, Cramer et al. introduced the use of diiodoethane 23 as linking agent, combined with tris-(2-carboxyethyl)phosphine (TCEP) as disulfide reductant for a one-pot reaction in aqueous environment.37 TCEP is a water-soluble phosphine-containing that breaks the disulfide bond by
performing a nucleophilic attack at one of the sulfides, creating a sulphur-phosphor bond and a thiolate anion (Scheme 6).38 Hydrolysis of the phosphor displaces the sulphur to create an inert phosphine
oxide and a second thiol after protonation. The application of TCEP allows reduction and modification naturally occurring proteins that contain disulfide bridges.
The research group optimized the one-pot procedure of reducing the disulfide bond to in situ create two free thiols, which undergo subsequent alkylation with diiodoethane 23,generating a stabilized thioacetal. It was noticed that a base was required for this reaction, presumably to induce hydrolysis. Optimization demonstrated that triethylamine (Et3N) is the most suitable base for this reaction. It was
reported that this method is completely chemoselective towards thiols, as other potential interfering amino acids were innocent bystanders.37
Scheme 6: Mechanism of TCEP-mediated disulfide reductions, followed by double alkylation by diiodoethane 23.
With the one-pot reaction optimized, stabilization of numerous peptides has been employed to yield SCS-peptides 24-28 (Scheme 7).
19
Scheme 7: The scope of methylene stabilized peptides.
Incubation of the methylene-stabilized and unstabilized with glutathione (GSH) validated that the stabilization method significantly increases the resistance towards disulfide reduction, displaying the potential of the methylene linker as application for therapeutics.
3.2 Oxetanes as cysteine linkers
A similar approach towards disulfide alkylation was demonstrated in 2017 by Bernardes et al., who used a one-pot alkylation method with TCEP and 3,3-bis(bromomethyl)oxetane 29 to install oxetane moieties into disulfide bonds (Scheme 8).20 Oxetanes are becoming a common motif in drug design, as
they can modulate features such as metabolic stability and solubility.39
This one-pot reaction is similar to the TCEP mediated alkylation introduced by Cramer et al., however, the method described by Bernardes utilizes potassium carbonate (K2CO3)instead of Et3N as base to
install oxetanes in disulfide bonds (Scheme 8).
20
It was demonstrated that this novel stabilization strategy has a wide reaction scope, as depicted in Figure 8. First, the oxetane moiety was inserted into disulfide bonds of biologically relevant peptides octreotide and somatostatin. These transformations led to a 4-fold increased bioactivity of oxetane stabilized somatostatin 30 towards the somatostatin receptor, while octreotide variant 31 does not display significant improvements.
Next, the oxetane insertion strategy had been applied on antibodies (Figure 8). Generally, transformations on large proteins like antibodies poses a greater challenge than on small peptides, as the disulfide bonds in antibodies are often sterically hindered.20 The first modified antibody was
Fab-Herceptin (Fab-Her), an antibody that is currently utilized in treatment of specific breast cancer types (Figure 8). This antibody contains a disulfide linker between the heavy and light chain. Unlike the native disulfide Fab-Her antibody, the oxetane stapled variant 32 exhibited high stability in reducing conditions.
Subsequently, the DesAb-Aβ3-9 antibody was stapled with the novel strategy. This antibody targets the
β-Amyloid peptide, which is a factor involved in Alzheimer’s disease.40 The stabilization of this antibody
proved to be a challenge, as the disulfide bridge is highly buried in the protein’s framework. A long reaction time was required and up to 40 equivalents of TCEP was needed for disulfide reduction. Nonetheless, oxetane insertion was successful to obtain protein 33, although no specific structural and bioactivity changes have been observed.
Lastly, carrier protein CRM197 had been treated by the oxetane insertion strategy to yield protein 34.
CRM197 is therapeutically important as it is exploited in multiple glycoconjugate vaccines.41
Regioselective stapling of this protein was successful and multiple bioassays have revealed that this novel protein contains enhanced immunogenicity in vivo.
21
Figure 8: Scope of the oxetane insertions into disulfide bonds, adapted and modified from Bernardes et al.20
Installing oxetane moieties by applying TCEP has shown great potential in the stabilization of disulfide containing peptides, as well as large biomolecules such as antibodies with sterically hindered disulfide bridges.
So far, novel strategies alkyl insertion into disulfides have been discussed. These cross-linkers display enhancement regarding stability and solubility, but do not allow follow-up chemistry to further modify the linker. The next example demonstrates a cross-linker that consists of a chemical handle for follow-up chemistry, which is useful for future bioactivity experiments.
3.3 Acetone-based cysteine linkers
Dawson et al. has reported the potential of acetone-linked peptides, both as peptide stabilizers and for follow-up chemistry where the acetone functional group is modified by different tags.42 This
22
the haloalkyl is dichloroacetone (DCA, 35) is linked between two cysteines or the non-proteinogenic amino acid homocysteine.
The cyclization method was applied on two peptide sequences, namely [Ac-YGGEAAREAXAREXAARE-CONH2], where X is either cysteine 1 or homocysteine 36 (Scheme 9). Previously, Greenbaum and
DeGrado reported the use of meta-dibromoxylene linkers (a 9-atom linker) on this peptide sequence, revealing greater helix stability than by using para (10-atoms) or ortho (8-atoms) linker variants.21 This
result is consistent with the results found by Dawson, where the homocysteine acetone-linked peptide (9-atoms) shows greater stability than the cysteine variant (7-atoms).
Subsequently, the homocysteine variant was further modified with oxime-based tags 38 on the acetone carbonyl, forming imine linkers 39. A total of eight oxime-based tags (38) were utilized. These tags consist of a wide range of functionalities, such as enhancing cell penetration using a polyarginine tail, allowing the use of fluorescent tags for bioassays or FLAG epitope tagging for immunological assays.43
Scheme 9: DCA 35 modifications of peptides 1 and 36, followed by tagging of the homocysteine carbonyl variant.
Lastly, macrocyclization was performed on peptides containing a single cysteine, allowing sulphur-nitrogen chain-tail linking with DCA 35. This reaction can be performed at the same conditions as the disulfide linking reaction (Scheme 10).
23
Scheme 10: Application of DCA 35 for tail-chain macrocyclization to obtain peptides 40 and 41.
Peptides 40 and 41, which contain a 16-membered ring had been synthesized, showing potential for DCA 35 as useful tool in tail-chain macrocyclization. Most of the examples include chain-chain connection rather than chain macrocyclization. The next example, however, includes a novel tail-chain connection strategy.
3.4 Chloroacetamide linkers for cysteine-tail macrocyclization
Suga and co-workers have described a method that does not rely on the use of SPPS, but by conducting genetic code reprogramming.44 Here, codons assigned for proteinogenic amino acids are reassigned to
non-proteinogenic amino acids, allowing synthesis of non-natural peptides.45 Macrocyclization is
achieved by the application of N-terminal chloroacetamides, followed by spontaneous macrocyclization between a cysteine residue and chloroacetamide.
Genetic code reprogramming was accomplished using Escherichia coli bacteria where certain amino acids are withdrawn from its environment, leaving a vacant site for novel amino acids. The targeted cyclic peptide was G7-18NATE 43, known as a potent tumor-growth inhibitor.46 The authors envisioned
that this macrocyclic peptide could be synthesized from its linear precursor Nα-ClAc-(G7-18NATE) 42,
which contains non-canonical amino acid N-chloroacetyl tryptophan (Scheme 11). Ribosomal synthesis by reprogramming yields the linear precursor which spontaneously cyclizes to obtain macrocyclic product 43.
24
Scheme 11: Genetic code reprogramming and subsequent tail-chain macrocyclization with chloroacetamides to form
peptide 43, adapted and modified from Suga et al.44
With the success of G7-18NATE X38 synthesis via in situ macrocyclization, it was decided to investigate the effect of ring size on the macrocyclization. Three peptides (44-46) were designed with arbitrarily chosen sizes (4, 6 and 14 residues, Figure 9). The sequences of these peptides were based on G7-18NATE 43, containing the crucial residues responsible for bioactivity. Furthermore, the peptides were tagged at the C-terminus with a Flag-peptide to ensure MALDI-TOF ionization for analysis.
Figure 9: Chloroacetamide-mediated macrocyclization products with varying ring sizes.
The successful synthesis of these three peptides suggests that the in situ macrocyclization is independent of the sequence and ring size. A mentioned limitation of this technology is that the sequence can only contain a single cysteine residue since multiple cysteines will lead to competing macrocyclizations, restricting the peptide scope.
Until now, examples of macrocyclized peptides are given which exhibit enhanced bioactivity upon cross-linking. However, other classes of constrained peptides display increased bioactivity after an external trigger. A new generation of cross-linkers increase the bioactivity after exposure to light with
25
a specific wavelength, acting as an on/off-switch. This allows cis-trans isomerization of functional moieties such as the azo-group. Cis-trans isomerization may contract the linking distance, altering the peptide’s conformation
3.5 Iodoalkyl-based photoswitch linkers targeting cysteines
One example of light responsive azo-containing peptide linkers is given by the group of Woolley.47 In
this research, molecular modelling was employed to design model peptide 47 with the sequence [Ac-EACARVAibAACEAAARQ-CONH2] (Aib=α-aminoisobutyric acid) (Scheme 12). Macrocyclization of the
two cysteine residues at i, i+7 position with azobenzene linker 48 yields photoswitch peptide 49. Exposing this peptide to light isomerizes the azo moiety from thermodynamically stable trans to cis, of which the latter stabilizes the peptide into α-helix conformation (determined by CD spectroscopy). As expected, the cis-conformation contains all the required hydrogen-bonds to stabilize the peptide backbone in α-helix conformation.
Scheme 12: Cross-linking of precursor 47 with photoswitch 48. Exposure to light with 380 nm wavelength results in
isomerization towards the cis conformation.
Further developments of azo-based photoswitch peptides towards bioactive compounds are mentioned by Heinis et al.48 This group utilize the phage display technique to encode random peptides,
followed by macrocyclization with photoswitch-linkers and exposure to UV light. The proteins were screened against model protein streptavidin to find potential bioactive hits.
Numerous examples of halobenzyls and haloalkyls as peptide stabilizers have been mentioned, covering the majority of sp3 hybridized linkers. However, many sp2-based reagents are reported as tool
for macrocyclization. The upcoming chapters will focus on the importance of sp2 hybridized linkers,
26
4 Conjugated cross-linkers and maleimides
4.1 Cysteine connected allene-based photoswitch linkers
Previously, peptide synthesis by applying SPPS and genetic code reprogramming were mentioned as methods to express peptides and proteins. A third approach is via phage display. Phage display is a technique where a protein encoding gene is placed in a bacteriophage (a virus that infects bacteria).49
This leads to expression of the corresponding protein on the outside of the phage, after which studies on the protein can be implemented. In many cases, screening of the protein-expressing phage is utilized to study protein-protein, protein-peptide or protein-DNA interactions.
This technique was applied by Derda et al., who combined phage display with peptide linkage.50 Similar
to the example of Woolley et al., the linker consists of a azo moiety that switches from trans to cis upon exposure of a certain wavelength of light. In this case, the linker contains two sulfonic groups, which significantly increases the hydrophilicity of the linker. Furthermore, the cross-linker is attached on cysteines via an allene functionality instead of an iodoalkyl. The novel linker, 3,3′-bis-
(Sulfonato)-4,4′-bis(buta-2,3-dienoylamido)azobenzene) (BSBDA 51) is based on the earlier reported
3,3'-bis(sulfonato)-4,4'-bis(chloroacetamido)azobenzene (BSBCA 50), which contains a chloroacetamide connecting group instead of an allene (Figure 10).51
Figure 10: Structures of photoswitches 50 and 51.
BSBDA 51 was subjected to phage displayed peptide sequence [ACPARSPLEC] (52). Since this peptide is synthesized by a living organism, the peptide contains a naturally occurring disulfide bridge. Prior to peptide macrocyclization, the peptide was reductively treated with TCEP to obtain two free thiols on which double thiol-conjugate addition took place, yielding peptide 53 (Scheme 13). Cis-trans isomerization was then achieved by irradiation with 365 nm LED light.
27
Scheme 13: Insertion of photoswitch 51 in peptide 52 to synthesize peptide 53.
It is mentioned that the novel photoswitch linker performs macrocyclization with a notable higher rate than previously reported BSBCA 50, demonstrating the potential of cross-linker 51as light-responsive linker. With the completion of the photoswitch macrocyclization on a phage displayed peptide, the authors mention that a library of these peptides can now be created, which in future work can be used for selection of bioactive ligands.
Photoswitch linkers have shown powerful post-cyclization functionalities such as secondary structure mediation for α-helices and enhancement of bioactivity. However, other post-cyclization functionalities also have emerged as powerful tools. One of these examples is reversible macrocyclization.
4.2 Meldrum’s acid derived linkers for cysteine-N-term linkage
Anslyn et al. introduced a Meldrum’s acid derived conjugate acceptor 54 as functional linker, which cyclizes the N-terminal with cysteine residues.52 In earlier work, it had been demonstrated that this
28
this reaction is reversible, as addition of DTT will ‘declick’ the cysteine sulphur and peptide nitrogen to regain the original reagents and the deactivated conjugate acceptor (55). This novel method was applied in the macrocyclization of a series of peptide sequences.
Scheme 14: The principle of ‘clicking’ and ‘declicking’ of conjugate acceptor 54.
All peptide macrocyclization were performed in presence of TCEP to prevent disulfide formation (Scheme 6). Initial cyclization studies revealed that at least 4 amino acids are required between the reactive sites, as smaller amounts would not cyclize cleanly. Further studies displayed that the linking was selective towards the N-terminus, while leaving amine residues such as lysine unreacted. The authors suggest that this is because of the difference in pKa of the amine moieties. The N-terminus is
less protonated than the lysine residues, making the N-terminus more prone to cyclization. A library consisting of 18 different peptides have been exposed to the novel cyclization method, of which 16 were successfully transformed (Scheme 15). Each peptide has been characterized by HPLC-MS analysis.
29
This macrocyclization strategy is compatible with all canonical amino acids and it is mentioned that future studies will attempt this method in presence of non-canonical residues. Furthermore, this methodology will be applied in future studies where the macrocyclized peptides will be screened for antimicrobial activities.
4.3 Macrocyclization and functionalization using divinyltriazines on cysteines
A third example of conjugate acceptor cross-linkers is given by Spring et al., who demonstrated the use of divinyltriazines (DVT 57) for macrocyclization via cysteines. This novel technique is applied to cross-link cysteines at the i, i+7 position to enhance the helical content of the peptide. Macrocyclization was first performed on the model sequence [Ac-ETFCDLWRLLCEN-NH2]56, of which the wild type
sequence was [Ac-ETFSDLWRLLPEN-NH2]. The wild type sequence targets p53/MDM2 proteins and
was used in previous work by the same group.54 Attempts to macrocyclize the model peptide required
screening of compatible solvent, revealing that a combination of sodium phosphate buffer with MeCN is ideal. Exposing random coiled sequence 56 to DVT 57 led to synthesis of helical peptide 58, determined by HPLC-MS and CD spectroscopy (Scheme 16).
To test the generality of this macrocyclization, biologically relevant peptides somatostatin, oxytocin and urotensin-II were successfully rebridged (Scheme 16). All cross-linking reactions were nearly complete in one hour.
30
Scheme 16: Macrocyclization with DVT 57 and the peptide scope.
To test the reductive stability of56compared to 58, a serum stability assay was performed. In all cases, the constrained peptide 58 showed significant stability compared to wild type 56. Additionally, competitive FP assays revealed that constrained peptide 58exhibits a 4-fold improved affinity towards MDM2 compared to the wild type linear sequence.
Linker 57enables further functionalization through the carboxylic functional group, allowing amide coupling of active groups such as fluorophores or cell penetrating peptides. The group of Spring modified linker 57with a nitrobenzofurazan (NBD) fluorescent tag, creating linker 62(Scheme 17). This linker has been installed in the model peptide 56 to synthesize peptide 63. It has previously been reported that NBD tags are utilized for generating fluorescent bioconjugated proteins.55 In contrast to
the initial coupling conditions, the cross-linking with 62requires modified conditions consisting of 7 : 1 : 1 sodium phosphate/acetonitrile (MeCN)/dimethyl sulfoxide (DMSO) (pH 8.0).
31
Scheme 17: Constraining peptide 56 with altered cross-linker 62.
As expected, the stapling with novel linker 62resulted in improved helicity compared to the linear sequence 56. Although this reported work does not include biological assays using peptide63or other NBD linked peptides, future studies are planned to perform these experiments.
A related cross-linker class are maleimides. The reactivity these moieties is similar to the latest three examples mentioned in this chapter, as maleimides are cyclic conjugate acceptors. Maleimides are popular in the field of bioconjugation due to their fast reaction rates and their additional functionalization potential.56
32
4.4 Bromo- and thiophenol maleimide linkage of cysteines
Baker and co-workers reported dibromo- and dithiomaleimides (64 and 65) as convenient reagents for disulfide macrocyclizations.57 Both reagents have been found after screening numerous maleimides on
the somatostatin (68) peptide. Macrocyclization was accomplished in a one-pot sequence where somatostatin’s disulfide bridge is reduced with TCEP to form two free thiols in situ. Subsequent conjugate additions of the thiols on the maleimides resulted in constrained maleimide constrained somatostatin 69 (Scheme 18).
With the successful novel synthesis of peptide 69, extended chemistry to build an extra functionality into the maleimide was envisioned. An N-poly(ethylene glycol) (PEG) maleimide variant was designed and synthesized as cross-linker for somatostatin (66) (Scheme 18). Incorporation of PEG groups on peptides increases its in vivo stability and is therefore a powerful method for measuring biological activity.58
Scheme 18: Transformation of somatostatin 68 with various dibromo- and dithiomaleimides.
Further bioactivity experiments of PEG-maleimide-somatostatin 70revealed that the peptide retained its bioactivity. It is mentioned that this novel strategy will broaden the scope for peptide modifications. Haddleton et al. utilized the same principle of installing a functionalized PEG labelled dithiomaleimide in the hormone oxytocin.59 Here, the only difference in methodology is a different connectivity of the
maleimide to the polymer, displaying the versatility of the macrocyclization method. While these 2 mentioned examples show insertion of functionalized maleimides into disulfides, a second approach to maleimide functionalization has also been reported.
4.5 Dibromomaleimide insertion into disulfides and post cyclization functionalization
Recently, dibromomaleimides (64) have been applied in modification of BH3 peptide BID (71).60 The
linker was placed at the i, i+4 position by utilizing TCEP in a one-pot transformation in previously reported reaction conditions (Scheme 19), yielding peptide72with enhanced helicity (determined by CD spectroscopy analysis) compared to its precursor. This work describes the first successful attempt to constrain a peptide into its α-helix conformation by employing a maleimide cross-linker.
33
Scheme 19: Constraining of peptide 71 with dibromomaleimide 64 to synthesize peptide 72.
The success of this modification inspired the researchers to expand the maleimide chemistry by functionalizing bromomaleimide 64. It was anticipated that the maleimide nitrogen could be provided with an alkyne handle for follow-up ‘click’ chemistry with azides. This was achieved in presence of diethyl azodicarboxylate (DEAD), triphenyl phosphine (PPh3) and propargyl alcohol, performing the
Mitsunobu reaction (Scheme 20).57 Activation of PPh
3 with DEAD is followed by attachment of
propargyl alcohol with PPh3. Subsequent nucleophilic attack of maleimide amine with the activated
propargyl alcohol yields alkyne-bromomaleimide73.
Scheme 20: Mechanism of the Mitsunobu reaction.
BID 71 was then treated with TCEP and alkyne-maleimide 73to obtain constrained peptide 74 (Scheme 21).
34
Scheme 21: Constraining peptide 71 with alkyne-functionalized maleimide 73 to obtain peptide 74.
Peptide 74was subjected to a series of azides to promote click chemistry, yielding additional triazole functionalized peptides 75 (Scheme 22).These functionalities include biotin (76), fluorescein (77) and PEG (78).
35
Scheme 22: Various ‘click’ reactions on alkyne-functionalized peptide 74.
It is displayed that the cross-linking is a reversible process, as the presence of glutathione allows the maleimide to be displaced, releasing the moiety and creating the non-cyclized peptide precursor. This characteristic, along with the stabilization of the α-helical motif and installation of additional functionalities via click chemistry, illustrates the potential of alkyne-dibromomaleimides 73.
To summarize, conjugated linkers may often contain post-macrocyclization functionalities such as light-mediated on/off switches, reversible constraining and additional tagging. Furthermore, most of these conjugate sp2-hybridized linkers have shown to be appropriate α-helix stabilizers.
A third class of sp2-hybridized linkers comprise haloaryls, on which nucleophilic aromatic substitutions
(SnArs) can take place. Due to the electron-withdrawing effects of the aryl substituents, the
cross-linkers are suitable for linkage between disulfides and amines at mild reaction conditions. The employment of haloaryls as peptide cross-linkers is described in more detail in the next chapter. It is
36
worth to mention that the lab of Bradley L. Pentelute is one of the largest contributors of this methodology, thus most of the examples mentioned in the next chapter are reported by this author.
37
5 Haloaryls as cross-linkers by nucleophilic aromatic substitution
5.1 Perfluoroaromatic monoaryls and bisaryls as disulfide linkers
Pentelute et al. demonstrated the use of perfluoroaryl in the macrocyclization of the relaxin H2 peptide
79.61 Targeting this peptide is of importance, as it has gained attention for its potential cardioprotective
ability.62 This peptide consists of an A- and B-chain, each encoding an α-helix conformation.
Furthermore, the two chains are linked to one another via two disulfide interchains and the A-chain contains an intrachain disulfide (Figure 11).
Figure 11: Structure of H2 relaxin 79.
Attempts are made to transform the peptide into a functional drug, however, low serum stability proves to be a major drawback.63 The hypothesis in this work was that stabilization of the
intramolecular disulfide bridge would improve the serum stability in reductive environments.
Synthesis was performed by creating the A- and chain separately through SPPS methodology. The B-chain required the utilization of native B-chain ligation (NCL) to obtain the free thiols in the peptide. Synthesis of the A-chain was done by implementing two strategically placed acetamide (Acm) protecting groups on cysteines (80, Scheme 23). Chain 80 was subsequently exposed to hexafluorobenzene 81to obtain peptide 82. Protecting group removal, followed by B-chain installation to finalize synthesis of constrained H2 relaxin 83.
38
Scheme 23: the synthesis strategy towards constrained peptide 83.
Although the peptide stapling was successfully performed, bioassays with this peptide revealed that the potency diminished significantly. It was earlier established that modifications on this disulfide site led to decreased affinity and the cause of this is still not clear.64 The authors mention that factors such
as linker rigidity, lipophilicity ad length might interfere with peptide-protein interaction and these parameters will be explored in future studies.
No regioselectivity issues occurred for the macrocyclization step and solely 1,4-thiol linkage was observed. It is mentioned by Pentelute that the first thiol SnAr creates the thiol-linked aryl, which
promotes a second thiol linking on the para position because of its stabilized intermediate (Scheme 24).65 Similarly, only cysteine arylation occurred and no side-products such as lysine or serine linked
peptides were observed.
39
In a different research, the same group also tried to explore effects of bisaryl linkage between disulfides.66 Model peptide 84 was synthesized, along with cysteine monoarylated analogue 85 and
extended bisarylated model 86(Scheme 25). By utilizing CD spectroscopy analysis, it was observed that model peptide X64 contains 16% helical content. Monoaryl-linked peptide 85contains an increased helical content of 53%, while bisaryl-linked analogue 86contains 36%.
Scheme 25: constraining of model peptide 84 with mono- and extended aryls.
The diminished trend in helical content for peptide 85 to analogue 86was not surprising, since the extended linker length is significantly longer than the distances of cysteines at the i, i+4 position in the model peptide.
A similar trend in regioselectivity of the extended linker was observed as for the monoaryl linker, where only 1,4-thiol linkage had taken place. Although a rationale for this regioselectivity was given for the monoaryl linker (Scheme 24), no explanation was presented for the extended linker selectivity. However, the chemical rationale is presumably the same in the case of both linkers.
In both mentioned cases, a chemoselectivity for the SnAr on haloaryls was observed for thiols. To
expand the chemical toolbox for peptide macrocyclization, the same research group attempted haloalkyl linkage between different nucleophilic residues.
5.2 Lysine-targeted N-arylation with haloaryls
In contrast to the extensively investigated cysteine arylations, reports of lysine N-arylations are still scarce. 55, 67, 68 Arylated cysteines may sometimes contain stability issues due to S-C(aryl) elimination,
40
forming dehydroalanine and destabilizing the peptide’s conformation.69 It is hypothesized that N-aryls
overcome the stability issues that are often present in S-aryl structures.
Typically, formation of N-C bond in these arylations forms a problem, as constructions of these bonds require high temperatures or the use of expensive metals. Pentelute and co-workers established a novel method of lysine N-arylation by utilizing bisaryl halides for lysine SnAr. This strategy is applied in
macrocyclization of biological relevant peptides.
During the screening of potential haloaryl linkers on a lysine-containing model peptide, two candidates emerged: perfluorosulphone 87 and dichlorotriazine-based 88(Figure 12). For these cross-linkings, (hydroxymethyl)aminomethane (Tris) or N, N-diisopropylethylamine (DIPEA) were used as base.
Figure 12: Structures of cross-linkers 87 and 88.
These two cross-linker were most reactive towards the lysines. To further determine the versatility of the cross-linkers, lysine arylation was performed in on a model peptide containing other potential competitive residues except cysteine. Macrocyclization resulted exclusively on lysines, confirming the lysine selectivity.
Additionally, the structural diversity of the macrocyclizations was explored by varying the lysine position (i, i+1 up to i, i+14) on model peptide 89 (Scheme 26).
41
Cross-linking on these variations worked at all lysine positions except i, i+9 and i, i+10. It is hypothesized that the cause for this is ring-strain, as double arylated products were observed. With the scope of the macrocyclic ring explored, the investigators then focused on the N-aryl chemical stability compared to
S-aryl constrained peptides, with 87as cross-linker (Figure 13). For this, model peptides90and lysine analogue 91 were exposed to oxidative and basic conditions (pH 10) and measured over time by utilizing HPLC-MS.
Figure 13: Model peptide 90 and its lysine variant 91 and their persistence in oxidative and basic conditions, graph modified
from Pentelute et al.67
This experiment validated the hypothesis that N-C(aryl) cross-linking provides a significant higher stability than S-C(aryl) in basic and oxidative environments. Based on these findings, the group modified an existing MDM2 inhibitor with N-arylation to display multiple desirable characteristics that may support cancer studies. Follow-up research focused on exploring the scope of thiol arylation along with strategy development.
5.3 Extended perfluoroaromatic aryls targeting cysteines
To further explore the possibilities of the haloaryl linkers, Pentelute designed an additional strategy to install extended perfluoraryl linkers between cysteines.65 This method involves a one- or two-step
approach of inserting the linker in two cysteines at different positions (i, i+1 up to i, i+14) of model peptide 92 (sequence=[IKFTNGLCCLYESKR], Scheme 27). The one-step approach comprises direct linking of the perfluoraryl linker into two thiols, while the two-step approach involves first S-arylation of both cysteines utilizing perfluoroaryl 81, followed by macrocyclization of the two aryls with commercially available dithiols (94).
42
Scheme 27: Two novel strategies towards cyclized peptides 95, schemes obtained and modified from Pentelute et al.65
Most of the linkers were installed by employing strategy I, obtaining yields greater than 60% at all positions. Additionally, strategy II was developed to circumvent steric hindrance that may occur while conducting strategy I. Rapid macrocyclization was observed when treating the arylated cysteines with diols, which is explained due to the smaller steric hindrance than by utilizing strategy I.
The authors mention that this novel methodology expands the macrocyclization toolbox for stabilizing peptides, allowing promising studies towards peptide-based therapeutics.
Fluoroaryls have shown potential as monoaryl or extended linkers on cysteine thiols and lysine amines, allowing stabilization and three-dimensional structure constraining. However, this class of linkers have not introduced additional post-macrocyclization functionalities such as fluorescent tagging or photoswitches. The next example demonstrates chloroaryls as cysteine linkers with additional functionalities.
43
5.4 S-tetrazines as reversible cysteine linkers
Smith et al. introduced dichlorotetrazine 96 as reagent to constrain disulfides with a S-tetrazine linker.70 Earlier research had demonstrated that the S-tetrazine moiety can be triggered by a
photochemical signal, releasing N2 and forming two thiocyanates.71 Here, this functionality is employed
to staple and photochemically unstaple peptides.
S-tetrazine was inserted in presence of monosodium phosphate in chloroform, resulting in 8 examples
of macrocyclized peptides 97 (Scheme 28). Irradiation with a UV lamp (312 or 365 nm) succeed in 5 examples to yield thiocyanates 98. Subsequent treatment of thiocyanates 98 with cysteine completed the unstapling, providing the initial peptides.
Scheme 28: Tetrazine-mediated stapling and unstapling of model peptides.
It is believed that the mechanism of cysteine mediated cyano removal is analogous to the native chain-linking mechanism (Scheme 29). Nucleophilic attack of the cysteine sulphur on the cyano carbon creates a thiourea. The cysteine amine performs a nucleophilic attack on the thiourea carbon, after which elimination of the thiol retrieves the original peptide while creating thiazolamine.
44
Tetrazines have a significant advantage, since these functional groups can be incorporated in the inverse electron demand Diels-Alder (DA) reaction.72 S-tetrazine undergoes a [4+2] cycloaddition,
followed by rearrangement with depletion of N2 to form a diazine (Scheme 30). This reaction is called
a reverse demand DA because regular DA reactions include [4+2] cycloaddition with an electron poor dienophile with a diene, while this reaction contains an electron poor diene (tetrazine) reacting with an electron rich dienophile (alkyne).
Constrained model peptide 100 was exploited with this reactivity, connecting alkyne fluorescein dye
99with S-tetrazine to synthesize macrocyclized fluorescein peptide 101(Scheme 30).
Scheme 30: The mechanism of the reverse-demand DA reaction and its application in the synthesis of a functionalized
peptide.
This novel method was applied in macrocyclization and fluorescein-linking of the thioredoxin protein, which was subjected to bioassays. This demonstrates the potential of the novel methodology. The authors mention that this chemistry provides novel opportunities for bioactivity studies.
To sum up, haloaryls prove to be important reagents for residue linkage via SnAr reactions. Fluoroaryls
exhibit diversity in the types of nucleophilic residues and much research have been conducted in the length of the linker. Fluoroaryls do not include additional functionality. Haloaryls such as dichlorotetrazine provide light-induced reversible peptide constraining and follow-up chemistry to attach fluorescein tags.
45
These examples demonstrate electron poor aryls to allow SnAr reactions on nucleophilic residues. A
different general method of heteroatom connection on aryls is via transition metal mediated cross-linking. This reactivity exhibits different synthesis potential, as will be discussed in the next chapter.
46
6 Palladium mediated cross-linking of aryls on residues
6.1 Palladium mediated arylation of cysteines
One of the pioneers of Pd-mediated cross-couplings towards heteroatom-C(sp2) bonds is Stephen L.
Buchwald, who is known for the Buchwald-Hartwig catalytic reaction to form N-C(sp2) bonds.73
Currently, Buchwald also focuses his work of transition metal mediated cross-linking for macrocyclization of peptides.
In 2015, Buchwald published a novel strategy, where Pd-aryl reagents are prepared and macrocyclized with cysteine residues.68 In contrast to Pentelute’s strategy of aryl activation by electron-withdrawing
halogens, Buchwald exploits solid storable activated [(RuPhos)Pd(X)]2Ar complexes ([Pd]2-aryl) to
promote reductive elimination with cysteine thiols.
Synthesis of the [Pd]2-aryl complexes was done by exposing dihalogenated aryls with
(cod)Pd(CH2TMS)2 in presence of RuPhos at room temperature (Scheme 31).74 The [Pd]2-aryl
complexes are stable and can be stored for a long time. This synthesis strategy allows [Pd]2-aryl
formation with a wide range of aryls.
Scheme 31: Synthesis of various Pd-reagents.
While tuning the reaction for the diarylation of cysteines, it was found that solvents MeCN:H2O (1:1)
with 0.1 M Tris (pH 7.5) at room temperature are the optimal conditions (Scheme 32).74 Quenching of
the reaction was done by addition of 3-mercaptopropionic acid.
These reaction conditions were utilized for aryl scope exploration performed on two model peptides (Scheme 32). Peptide 102(H2N-ILTFCDLLCYYGKKK-CONH2) has two cysteines with position i, i+4, while
peptide 103 (H2N-LTFCHYWAQLCS-CONH2) contains cysteines at i, i+7 position. A total of 8 aryls have
47
Scheme 32: The Pd-mediated macrocyclization approach and the reaction scope.
Macrocyclization was performed with yields ranging from 70 to 99% on either the i, i+4 and i, i+7 position. Although no mechanistic studies have been performed, it is believed that the S-C(aryl) bond is formed via reductive elimination (Scheme 33).
Scheme 33: Mechanistic rationale of the Pd-mediated arylation of cysteines.
Because of the mild reaction conditions, wide scope, high yields, cysteine position scope and facile [Pd]2-aryl reagent synthesis, this novel cysteine arylation methodology is considered very versatile.
As the Pd-mediated arylation of cysteines has been accomplished, the focus had then switched to N-arylation of lysines. Compared to cysteine constraining, lysine ligation forms different challenges as the reactivity of primary amines is different from thiols.
6.2 Palladium mediated arylation of lysines
In contrast to cysteine conjugation, it was envisioned that lysine modification would require a different strategy to address the lower nucleophilicity of amines and lower acidity of the Pd-amine complex.75
