Template assisted synthesis of a planar chiral pre[2]rotaxane

(1)

Template assisted synthesis

of a planar chiral

pre[2]rotaxane

Master Thesis

By

Tom Besseling

(11873639)

February 2020 – January 2021

MSc Chemistry

Molecular Sciences

48 EC

Universiteit van Amsterdam

Van ‘t Hoff Institute for Molecular Sciences

Synthetic Organic Chemistry Group

Examiner: Prof. Dr. J.H. van Maarseveen

Second Examiner: Prof. Dr. J.N.H Reek

Daily supervisor: Simone Pilon MSc.

(2)

(3)

Abstract

Since the synthesis of the first ever rotaxane by Harrison and Harrison, the search for these kind of molecular interlocked molecules has advanced significantly. In this area, rotaxanes synthesis is widespread. Exciting compounds have been synthesized and multiple applications have been discovered for these new compounds. In the search for more complex applications, like the

controlled synthesis of lassopeptides, that can be seen as chiral [1]rotaxanes, the control of chirality in rotaxanes will become more and more important. The same counts, for instance, for the chiral rotaxane, published by Goldup in 2020, used as a catalyst. This control of chirality will also make it possible to, for instance, synthesize stereospecific [3]rotaxanes. Those compounds will have a completely new set of applications like the use as a bidentate ligand.

The aim of this project was to make the desired planar chiral pre[2]rotaxane in order to find new methods to possibly synthesize lassopeptides stereospecificly in the near future. Next to that, It was interesting to obtain this chiral pre[2]rotaxane, because of its two enantiomers. If we are able to separate these enantiomers by prep HPLC, we could specifically manage the chirality while forming a chiral [3]rotaxane. In this compound, both macrocycles will then have their directionality counter rotational with respect to each other. As third, this desired compound would be the first ever planar chiral pre[2]rotaxane, since all methods till now, were only able to obtain a chiral product when the rotaxane was already formed. To obtain the desired compound, both the ring fragment as the template had to be de-symmetrized. After successfully obtaining both compounds, the

pre[2]rotaxane could be made via CuAAC macrocyclizations. The final product was obtained after a 8-step synthesis. Some experiments were done, showing possible rotation of the inner template inside the larger ring. A new, sterically larger stopper has to be bound to prevent this from happening. Binding this before the macrocyclization might be the best step.

(4)

Abbreviations

4ÅMS: 4 Ångstrom molecular sieves

AcOH: Acetic acid

CuAAC: Copper catalyzed azide alkyne cycloaddition DBU: 1,8-diazabicyclo(5.4.0)undec-7-ene DCM: Dichloromethane DIPEA: N,N-Diisopropylethylamine DMAP: 4-dimethylaminopyridine DMF: Dimethylformamide DMP: 2,2-dimethoxypropane DMSO: Dimethylsulfoxide Et2NH: diethylamine Et3N: triethylamine

EtOAc: Ethyl acetate

HBTU: 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate HRMS: High resolution mass spectrometry

IR: Infrared Spectroscopy

LCMS: Liquid chromatography-mass spectrometry

LDA: Lithiumdi-isopropylamide

LiBH4: Lithium borohydride

MeCN: Acetonitrile

MeOH: Methanol

MOM: methoxymethylene

MOMCl: Chloromethyl methyl ether MsCl: Methyl sulfonylchloride

NBS: N-Bromosuccinimide

NCS: N-Chlorosuccinimide

NMR: Nuclear magnetic resonance

OMs: Methylsulfunate

On: overnight

PFPOH: pentafluorophenol

PPh3: triphenylphosphine

p-TsOH: para-toluenesulfonic acid

rt: Room temperature

TBTA: Tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine

THF: Tetrahydrofuran

TMS: Trimethylsilyl

(6)

Introduction

Mechanically interlocked molecules

Over the last couple of decades, research into interlocked molecules has been a hot topic. Multiple new forms of these compounds were made. In the name

of these new compounds, the prefix [n] points towards the amount of components that make up the

mechanically interlocked molecule. Some examples of these new compounds are for instance, the [2]solomon link (i) and the [3]catenane (ii) (Goldup1), or the

[1]molecular knot (iii), as described by Leigh et al2. Next to this, a main group of mechanically interlocked molecules are the rotaxanes.

Since the first ever synthesis of a [2]rotaxane was achieved by Harrison and Harrison3 (1967, Scheme 1), the search for rotaxanes in both organic chemistry as the supramolecular world has advanced significantly. The publishing of this first rotaxane relied solely on the statistical anomaly of rotaxane formation. The synthesis of this rotaxane was performed by binding a cyclotiacontanone on a Merrifield resin. The subsequent treatment with 70 cycles of 1-10-decandiol and trityl chloride gave only 6% yield of a [2]rotaxane.

Through the years multiple new rotaxanes were obtained. The yield of these new compounds and the complexity of these interlocked molecules have increased. This was possible, due to the development of new reliable methods, which can all be placed in one of the three main

preorganization methods. This preorganization makes use of interactions between two components, obtaining the desired form of the compound, before completing the last covalent bond. These three methods are; 1) capping, 2) clipping, 3) active metal template (Figure 24). In the capping approach, a linear thread is (mostly) supramolecular bound, inside of a macrocycle. Thereafter, a stopper is connected on each side of this thread. This prevents the rotaxane from dethreading. Due to these stoppers, the rotaxane is a covalent compound instead of a supramolecular one, since at least one covalent bond has to be broken, before both components can fall apart.

iii.

Scheme 1: Synthesis of the first [2]rotaxane by Harrison and Harrison3. After treatment with 70 cycles of 1,10-decanediol and trityl chrloride, the desired compound was obtained with a yield of just 6%. This means a yield of 0.08% per cycle.

Figure 1: Mechanically interlock molecules. i) Solomon

(7)

In the clipping approach, the thread is already stoppered, before the pre-macrocycle is formed around it. This formation is (mostly) enhanced by some form of supramolecular interactions. After this preorganization, the macrocycle is closed by adding the last part of this cycle. This method is also mainly used in the formation of catenanes. In the third approach, a metal-complex is coordinated to a motive, built in the macrocycle. This metal will then take the two thread fragments and react them inside of the ring. An adaptation to this method is the use of an organic template compound. This way, the template is bound to the ring fragments first, before the macrocycle is formed. A second adaptation is the synthesis of the thread inside of the ring due to interactions, instead of binding the metal-complex to the cycle first.

Thanks to the increasing complexity of the rotaxanes, more and more applications are possible as well. An example of these applications was published by Goldup et al. in 20205. They reported a new [2]rotaxane, which can be used as a stereospecific catalyst. This rotaxane was synthesized via the active template method. They used a typical Goldup

macrocycle, wherein they formed a thread by performing a CuAAC click reaction, ending up with a triazole in the centre of the thread.

To this rotaxane, a gold-complex was attached. This way, a gold-mediated variant of the Ohe-Uemura cyclopropanation of alkenes and propargylic esters was performed. Despite the fact that this rotaxane was not explicitely designed or optimized at the time for these kind of reactions, still stereoselectivities with benzoate esters of 45-75% ee were achieved. This was similar to for instance (R)-DTBM-SEGPhos(AuCl)2 (68% ee),

which was already optimized and widespreadly used for these reactions. This suggests unexplored potential of rotaxanes in catalytic applications.

Figure 2: Mechanisms of rotaxane synthesis4.

A

B

Figure 3: gold-mediated variant on the Ohe-Uemura cyclopropanation. A) Gold-coordinated [2]rotaxane. B) Catalysed Ohe-Uemura

cyclopropanation ending up with 45-75% ee.

(8)

Another great example of more complex rotaxanes was published by Stoddart and Bruns in 20146. They published 2 connected rotaxanes (see Scheme 2), the so-called [c2]Daisy Chain, that are able to contract or extent under pH-triggered conditions, mimicking the action of a human muscle. Those rotaxanes are polymerized using p-xylyl linkers. However, the average chain length they obtained was only of 11 repeating units far from an original human muscle. Under acidic conditions, those daisy chain rotaxanes extend, while they contract under basic conditions.

This proposed contraction/extension idea was confirmed by estimations based on measurements in

silico by the authors. They however, measured some of the values experimentally by performing

force measurements on individual [4]rotaxanes with atomic force microscopy. They measured the length of the rotaxane in the extended state as well as in the contracted state. This showed a contraction ratio between 40-55% of pH-stimulated rotaxanes. This is one example of the increasingly complex rotaxanes.

To keep increasing the complexity, and thereby the amount of applications, the translation to [n]rotaxanes had to be made as well. The previous example already makes use of a connected [4]rotaxane. However, much more [n]rotaxanes were published over the last couple of years. Multiple new methods to synthesize these [n]rotaxanes were found in as well organic as in

supramolecular chemistry. The first example is a supramolecular machine by Stoddart7,8 as well. They were inspired by carrier proteins in nature. These proteins consume ATP in order to pump ions or molecules across cell membranes.

In this machine, a positively charged ring was kinetically trapped onto a thread, by using externally driven oscillations of the redox potential. This tetra cation ring (CBPQT4+) can pass the first 3,5-dimethylpyridinium (PY+) group only under reducing conditions.

Scheme 2: Muscle-like compound, published by Stoddart et al. contracting and extracting under pH influences.

Figure 4: Kinetic diagram describing the assembly of a ring onto the collecting chain when driven by external oscillation of the redox potential7,8. Under reducing conditions, an equilibrium between Dred and Ired arises. Under oxidizing conditions, the

system equilibrates between Iox and Aox. Continual cycling between oxidizing and reducing condition pumps several rings onto

(9)

There it is stabilized by a viologen (V2+) recognition site. This of course is an equilibrium. Under oxidising conditions, removal of the ring from the thread is made impossible. The only way to go for the cycle is by passing the second obstacle, which is an isopropylphenylene (IPP) group. By employing reducing conditions again, the ring is trapped at the final strand, while a second ring can pass the first obstacle again. By oxidising, the only option is an equilibrium between the viologen site and the final strand for the second ring, which traps the first ring definitely at the final strand. By changing the conditions a number of times, the authors described to have obtained a maximum of 4 rings on the final strand.

A second good example to synthesize a [n]rotaxane is an organic method, reported by the Van Maarseveen group. In 20209, they published an article, in which 2 rings and 2 stoppers were

synthesized. They connected multiple rings by using linkers. In this paper they published a series of products containing [2]-, [3] and [4]rotaxanes. The macrocycles were formed around an organic template. The used linkers on this template had a terminal alkyne or a terminal azide. This way, 2 pre-rotaxanes could be bound via a CuAAC click reaction. The same counts for the stoppers. The 2 formed stoppers also contained an alkyne or an azide. This way, [3]- and [4]rotaxane compounds could be formed

However, due to the all carbon ring, and the use of the same stoppers on each side, these products are achiral. These examples already show the higher complexity. However, to go further, also chirality has to make its entrance in these kind of mechanically interlocked molecules. This was already seen in the before mentioned example from Goldup et al, where the Ohe-Uemura cyclopropanation was already making use of an enantiopure rotaxane.

Figure 5: [3]-, and [4]rotaxanes, published by Van Maarseveen et al9.

(10)

Chirality in rotaxanes

Making these kind of compounds chiral, would lead into a completely new species of compounds, with a lot of new possibilities and advantages. To make these new kind of compounds, ways to obtain this chirality had to be found. Therefore, two major methods were described by Evans in 201810. The two methods he wrote about were chirality arising from classical chiral elements and chirality arising from mechanical bonding. In this first method, the straightforward, “classical” way to create a chiral compound, was used by adding a chiral element. This chiral element can be added onto to the used stopper or onto the cycle. One of the first examples of this method was published by Vögtle et al. in 199711, where they used chiral tetra-acetyl glucose stoppers.

The second method is a bit more elaborate. In these compounds, chirality can arise even when interlocked components themselves are achiral. This way, because of the mechanical bond, more opportunities are possible. This can be described as “Mechanical chirality” which has been defined by Bruns and Stoddart as “A non-classical form of chirality resulting from the spatial arrangements of

component parts, connected by mechanical bonds”. The easiest way to obtain this, is by making use

of a topological prochiral ring. These rings have a pro-S and a pro-R face. This can be induced by introduction of certain functional groups inside the ring, like amides, esters or triazoles. Due to these functional groups, the cycle becomes directional (topologically prochiral). Also a directional thread can be used. With this method as well, Vögtle12 published one of the first examples. In 1997 he published a rotaxane containing a directional macrocycle, as well as a directional axle, both induced by the use of a sulfonamide. Unfortunately, they were unable at the time to assign the absolute configuration of the enantiomers.

In 2014, Goldup13 published an mechanically chiral rotaxane, in which he nicely showed both methods (Scheme 3).

Scheme 3: Method, reported by Goldup in 2014, to obtain enantiopure samples of the desired rotaxane.

(11)

This was actually a breakthrough in the search for methods to obtain enantiopure mechanically chiral rotaxanes. As can be seen in Scheme 3 , they used both a “classical” chiral element, by using a chiral stopper, and, as well as a topologically prochiral ring, with O-C-N directionality, as a directional thread, due to the triazole. Because the molecule possess both types of chirality, they would obtain diastereomers instead of enantiomers first. Due to a CuAAC “click”-active metal template synthesis, they obtained a 1:1 ratio of both diastereomers. Hereafter, they substituted the chiral part of the stopper with an achiral part, ending up with enantiopure compounds. By obtaining crystal structures of the diastereomers, they were able to assign the absolute configurations of these diastereomers, and therefore also of the resulting enantiopure samples.

The example of Goldup shows, what has been published by multiple others as well, that the usual method to obtain enantiopure samples of a rotaxane, is the synthesis of

diastereomers, followed by the substitution of the chiral part. This way, they already have the rotaxane, before getting the enantiomers. The same method was published by Takata14 in 2020. Other ways, like the Leigh published in 202015, using an activated N-acylation template, also shows the synthesis of the rotaxanes immediately. This way, a template-coupled planar chiral pre[2]rotaxane has never been synthesized before.

Through the years, applications for these enantiopure rotaxanes have been found as well. For instance, as asymmetric catalyst. In 2014, Leigh et al.16 published a pH-sensitive rotaxane (Scheme 4). Under acidic conditions, the catalytic reactivity is turned off, due to the placement of the crown ether cycle onto the amine in the thread. By

deprotonating this amine, the cycle can freely move over the thread, making the amine available for the substrate. Under these conditions, the rotaxane can catalyse Michael additions with reasonable conversion (70%) and high enantiomeric ratios of products (er = 94:6).

Next to this example, multiple catalytic applications have been found since then, as, for instance, the Ohe-Uemura cyclopropanation, described before.

The importance of chirality in these kind of compounds can be explained by two possible applications. The first application of these new methods might be used in the synthesis of lassopeptides. Since the first lassopeptide was discovered in 199117, 38 variants have been described till now. These lassopeptides are formed when an enzyme in a sequence folds the N-terminus of a peptide chain around itself and forms an amide bond with the side chain of either glutamic acid or aspartic acid. This way, a linear peptide is threaded through the cyclic part of the same sequence.

Scheme 4: Rotaxane catalyzed reaction of the Michael addition with 94:6 er. As published by Leigh et al.16 in 2014.

Figure 6: Structure and amino acid sequence of Microcin J25.18

(12)

This can be seen in Figure 6 where a typical lassopeptide18 is shown. This lassopeptide is microcin J25, that clearly shows an eight-amino-acid membered ring with a 13-amino-acid membered chain

through the cycle. As can be seen the N-terminus of the glycine amino acid, forms an amide bond with the side group of glutamic acid, after which the chain is sterically held in place by the side groups of phenylalanine and tyrosine.

As can be seen in Scheme 5, a lassopeptide contains a cyclic structure. Due to the chain that is threaded through this ring, these kind of compounds can be seen as [1]rotaxanes. This threading can occur from both sides of the ring. Since the sequence has a N-C-directionality, due to the coupling of amino acids, this means that the lassopeptide contains mechanical chirality. This means that the threading of this molecule, from both sides, give the other enantiomer. In the search to control this chirality, new methods can help.

A second possible application that needs control of chirality is the synthesis of an enantiopure [3]rotaxane.

As can be seen in Figure 7, to obtain a chiral compound, the directionality of both macrocycles need to be counter-wise from each other. If not, a meso-structure will be formed instead. A possible application for these kind of structures might be the use as a bidentate ligand. If a metal ion can bind to the centre of the two cycles, like it can to the centre of a [2]rotaxane, these kind of compounds might be able to stereospecifically catalyse reactions.

Scheme 5: Two enantiomers of a lassopeptide, due to the N-C-directionality of the sequence.

Figure 7: chiral [3]rotaxane. Since both macrocycles have their directionality counter-wise to each other, this makes the compound chiral.

(13)

Research goal & synthetic plan

In search for new methods to control mechanical chirality of lassopeptides and the possible synthesis of a stereospecificly chiral [3]rotaxane, the goal of this research project is the synthesis of this planar chiral pre[2]rotaxane (Figure 8).

It was decided to introduce two directional triazole groups along the macrocycle. Due to these two groups, a specific C-N directionality will be induced, which makes this macrocycle a topologically prochiral ring. These two triazoles are formed via a CuAAC click cyclization of two ring fragment molecules, around a central terephthalic

template. To end up with the two desired enantiomers, due to the planar chirality, to this central template, two different protecting groups are added.

A second advantage of these two different groups, is the possibility to separately cleave both protection groups, adding two separate stoppers before cleaving the template. This way, it’s also a possibility to add a stopper to one side and a connecting part to the other. While reacting these connecting parts, a stereospecific pre[3]rotaxane could be obtained as well (Figure 9). This desired compound 17 will be the first planar chiral pre-rotaxane. This beacause, in the contrary of the previously mentioned articles of Goldup and Takata where they made diastereomers of the rotaxane first and made the

enantiomers afterwards, this method of synthesizing an enantiopure pre[2]rotaxane isn’t used before.

As depicted in Scheme 6, the desired compound will obtained via a CuAAC Click cyclization of coupling compound A2. This coupling compound is the actual product of two ring fragment

molecules (A3) and one template molecule. This can be done by making a PFP-ester of B1. This can be synthesized by the addition of two protection groups on the oxidated form of the commercially available dimethyl-2,5-dioxocyclohexane-1,4-dicarboxylate. The ring fragment can be synthesized out of A4 via two alkoxide substitutions on the mesylates. This compound will be obtained by a bis-hydroxymethylation from the commercially available 4-methoxyphenol.

Figure 8: Desired pre[2]rotaxane

(14)

Next to this synthetic plan, a route that was started already before, was also worked on during this project. The aim of this research route was exactly the same, ending up with a planar chiral ring, containing two triazole groups. In addition to that, the macrocycle would also contain two secondary amides, to introduce four C-N-directional groups, instead of two. The same asymmetric terephtalic template will be used during this synthesis. As can be seen in Scheme 7, compound C1 again can be formed via a click cyclization of coupling product C2. This can be obtained by a coupling reaction of the previously described asymmetric template and ring fragment C3. This will be synthesized out of

C4, which will be obtained from C5 via a Heck alkynylation. C5 will be obtained out of the

commercially available 4-tert-butylphenol via a bis-hydroxymethylation and a chlorination with NCS.

Scheme 6: Synthetic strategy to obtain the desired pre[2]rotaxane.

Scheme 7: Second synthetic strategy. The adapted strategy to finally obtain a racemic mixture of a C-N directional macrocycle with two different sidegroups.

(15)

Results

Synthesis of the ring fragment.

Before the pre[2]rotaxane could be built, first the ring fragment and the asymmetric template had to be synthesised. The synthesis of this ring fragment was the first aim. As shown in the synthetic strategy, this compound was made out of 4-methoxyphenol. This was first bis-hydroxymethylated into compound 18. The second step was the protection of the phenol, to make sure, only the benzylic alcohols would react in the step afterwards. This protection was performed by addition of an allyl-group.

After this protection, the benzylic alcohols had to be substituted into good leaving groups. The first idea was a synthesis according to a procedure, reported by Cram and Lam19 in 1985. In this

procedure, the starting material is solved in cooled THF, together with PPh3. N-bromosuccinimide

(NBS), solved in THF, was then added to the reaction. The 1H-NMR showed product was obtained. However, it also showed side products, that could not be explained with the reaction mechanism. During literature search, a side reaction of NBS with THF was proposed, due to the possible presence of peroxides in the THF, which initiated a radical chain reaction with NBS20,21.

Scheme 9: Third step in the synthesis of the ring fragment. The reaction with NBS in THF gave side products and just 65% yield. The reaction with chloromethyl sulfonyl in DCM gave a much higher yield without any side products.

After this was found, a new idea came up, using a mesylate as the leaving group. This had been used before in this research group, and gave much higher yields of a pure product. As this product wasn’t stable enough to purify by column, it had to be synthesized pure. This was obtained by stirring the reaction at -10oC for a couple of hours. In the next steps, both leaving groups had to be substituted with a different substituents.

Scheme 8: First two steps in the synthesis of the ring fragment. In the first step a bis-hydroxymethylation is performed, secondly the phenol is protected by an allyl-group

(16)

To achieve this, first a reaction was proposed with the alkyne-substituent. This was done in a 1 to 1 equivalent reaction between the starting material and the substituent. Due to the sodium hydride, an alkoxide was made first, where after this was added to a solution of starting material. This gave a mixture of leftover starting material, the mono-substituted product and the di-product, according to the 1H-NMR. However, this product was still not stable enough to purify by column, since it

completely degraded over the column. This meant, the two steps had to be done both, before purification could be done. At the same time, it was proposed to start with the azido-alcohol substituent, since it had to be re-synthesized if it was used completely, while the alkyne-substituent was commercially available. By using the azido-alcohol compound first, only one equivalent would be used every synthesis instead of three equivalents. For this project, this compound was already available on the lab.

The first reaction was performed, ending up with the expected mixture of compound 22, the di-substituted compound and starting material. After a short work-up to remove leftover azido-alcohol compound, the second reaction was started the same way. This gave again a mixture of three products, the di-azide, the di-alkyne and desired compound 23. However, this time, they were easily purified by column.

To finally obtain the ring fragment, the allyl protection group had to be removed. This was tried first with a palladium(0)-complex, (Pd(PPh3)4), and phenylsilane. This gave a low yield as seen in Table 1,

Entry A. Phenylsilane was replaced by di-ethyl amine, which gave even lower yields. Multiple other conditions have been used. Even LDA was tried, to transform the allyl group into a vinyl group. This could then simply be removed with dilute hydrochloric acid. This gave no conversion at all. In the end, only replacing phenylsilane with potassium carbonate seemed to be working best. This is most likely because the reaction with a Pd0-complex needs a scavenger to intercept the formed cationic π-allyl complex. This is normal wise a compound from nucleophilic nature (amine / alcohol), or an hydride donor (like Et3SiH)22. However, with this reaction a base seemed to work better.

Scheme 10: Desymmetrization of the ring fragment by adding 1 equivalent of azido-alcohol compound, followed by an excess of 3-butyn-1-ol. As described, these two reactions were performed almost in situ, due to the instability of the mesylate.

Scheme 11: Deprotection of the phenol. Several catalysts, reagents and conditions were tried to remove the protecting allyl-group, as can be seen in the entries of Table 1.

(17)

Entry

Catalyst

Reagent

Solvent

Yield (%)

A

Pd(PPh

3

)

4

Phenylsilane

DCM

37 B

Pd(PPh

3

)

4

Et

2

NH

DCM

19 C

23

Pd/C

TsOH

MeOH + H

2

O

No conversion

D

24

-

LiBH

4

THF

No conversion

E

25

PdCl

2

+ CuCl

2

MeOH

No conversion

F

26

-

LDA

THF

No conversion

G

27

Pd(PPh

3

)

4

K

2

CO

3

MeOH

71

Table 1: Different conditions, under which the deprotection of the allyl group was performed. Unfortunately most conditions gave no conversion at all.

This way the desired ring fragment was made with only one low-yielding step. However, this was expected, since the compound had to be de-symmetrised during that step. Different strategies to desymmetrise the ring fragment, like using a acetonide on compound 18, to react first the only free benzylic alcohol and secondly, after removing the acetonide, the second benzylic alcohol, wouldn’t give higher overall yields, due to the increased number of steps. Other protection groups like a methoxy-group or chloromethyl methyl ether (MOMCl), wouldn’t help either, since a methoxy group needs harsh conditions to deprotect, and the MOM-group gives difficulties in managing the reactivity of the intermediates. The same desymmetrization problem would most likely be the case in the template synthesis as well, which will be shown in the next part.

(18)

The synthesis of the asymmetric template.

The synthesis of the asymmetric template was, already from the start, an interesting part of this thesis. This was mainly because of the symmetric starting material. The objective was to obtain the desired product with two different side groups, with an acceptable yield. At first two different approaches were performed. The first approach was published by Perron et al.28 in 2018. In their procedure, they used one equivalent of starting material and allyl bromide and 1.5 equivalent of potassium carbonate as base. The second procedure was published by Lee et al.29 in 2003. They used one equivalent of base and allyl bromide on four equivalents on starting material.

During these reactions, a mixture of the mono-allylated product and the di-allylated product was obtained, together with recovered starting material. The mixture had to be purified by column chromatography. The TLC already gave an unexpected result, having the starting material running as highest and the

di-allylated compound the lowest. This was the other way around as expected by polarity. Most likely, this was because of the ester on the ortho position. The phenol and the oxygen atom are likely to form hydrogen bonds, making this compound less polar. Unfortunately, purifying by column chromatography was impossible to do, since the three different fractions came of almost together. A second disadvantage was the poor solubility of the starting material in ethyl acetate, which made it crash out on the column, giving small amounts of starting material in every fraction.

In the end, no satisfying results were obtained since the yield of Lee’s procedure

was very low, due to the amount of starting material, and Perron’s procedure ended up with an one to one mixture of the mono and di-allylated product, with 25% starting material left. Other bases like KHCO3 and DBU were tried as well. Unfortunately, they gave no conversion at all.

Start material (2H) 7.49 ppm Di-allylated product (2H) 7.42 ppm Mono-allylated product (2x1H) 7.37 + 7.41 ppm

Figure 11: 1H-NMR of the product mixture, according to Perron's procedure.

Figure 10: Hydrogen-bond that makes the compounds less polar as expected Scheme 12: First two steps in the synthesis of the assymetric template. By adding 1 equivalent of

(19)

As a result, some adaptations were made on the procedure of Lee, reducing the amount of equivalents of the starting material. Next to this, addition of the base and allyl bromide was

alternately performed portion / dropwise. This showed that adding the allyl bromide dropwise to the reaction mixture, gave less of the di-allylated compound. Portion wise addition of the base made no difference. Unfortunately, the yield stayed really low.

After numerous small scale reactions, it was decided that in the end, the method with the highest yield and the lowest amount of leftover starting material and formed di-allylated product would be used, despite the low yield. This was performed on a 1g scale. During the work-up of this reaction, it was found that the di-allylated product was a solid, like the starting material, but the mono-allylated product was an oil. This way the purification could be tried by pouring the reaction mixture into a large excess of water, crashing out the starting material and adding a bit of ethyl acetate to obtain the desired product. Unfortunately, a mixture of both products and the starting material was still obtained. This mixture was than solved in a bit of EtOAc again, and filtered. This helped to get the product pure. However, this was obtained with a yield of 5%.

Since nothing seemed to work, a new option was proposed. The use of solid-phase organic chemistry. This makes use of a polymer resin, loaded with linking groups that act as protecting groups, synthesising the product step-by-step30,31,32. A resin with a bromide linker was found as seen in Scheme 1333. This resin, 2-(4-Bromomethylphenoxy)ethyl polystyrene, had a loading of 1,27 mmol/g. This was the lowest loading, that could have been used, to obtain an acceptable amount of product each time (250 mg starting material / g resin).

During these reactions, symmetric compound 25 was loaded onto the resin under the same conditions as the addition of allyl bromide, as seen in Scheme 13. In the second reaction, allyl bromide was added, under the same conditions again. After this reaction, the template could be cleaved with TFA. However, after cleaving the template, the 1H-NMR showed no signs of an allyl-group. Only starting material was obtained. This meant that both phenols were loaded onto a different linker molecule. This meant that the loading was most likely still too high. However, using a lower loading, wouldn’t give enough yield for this reaction. This method was abandoned.

At that moment, the most promising method was used again. This time, the focus was put on the work-up method. Since we already knew that the starting material was insoluble in almost every apolar solvent, and the solubility in ethyl acetate was low, this gave the opportunity to wash out the desired product together with the di-allylated product.

(20)

This was done on a large scale immediately. 4 grams of starting material were used. After stirring overnight, the work-up was done, finishing with a washing step of the product mixture with PE:EtOAc (10:1). 2.5g starting material was recovered (60%). A 1H-NMR was measured, which showed no more starting material in the product fraction. The obtained mixture (1.5g, 34%) was used in the second reaction with MOMCl. After the purification of this reaction, 685mg of the di-allylated product was obtained. This didn’t react during this reaction, meaning the starting mixture contained around 800mg of the desired product 26. In the end compound 27 was obtained with a yield of around 80% Now the template was de-symmetrised, it was necessary to make the template more reactive towards the phenol of the ring fragment. Therefore, the methyl ester had to by hydrolysed, followed by making a new ester with pentafluorophenol (PFPOH). This PFPOH, in the presence of a base, esterifies easily with the acid.

In the first step, the methyl ester was solved in a reaction mixture of four equivalents of potassium hydroxide in THF, methanol and water. This reaction was successful with a yield of 93%.

Unfortunately, the 1H-NMR showed three different sets of aromatic peaks. However, the integral of these peaks together, correspond to the integral for among others the internal allyl peak (6.10 ppm). The idea that arose first, was the possibility of three different protonation products. However, this was refuted by the fact that the product peaks disappeared, when the compound was in the acidic phase for too long / acidified slowly with 1M HCl. This can be seen in Figure 12. It shows the presence of all three sets of aromatic peaks in the first spectrum. In the second spectrum, it shows the

disappearance of the product peaks and the increase of both other sets of peaks. In addition to this, a second peak of both MOM-peaks appeared. This might show some rearrangement from the MOM as well. When this mixture was used in the next step, no reaction occurred. This showed that the two peaks at 8.11 and 7.89 are the set of peaks corresponding to the free acid.

The idea that arose next, was the possible decomposition at a certain pH / time in the acidic phase. This could be avoided, by acidification of the compound with concentrated HCl and immediate extraction with ethyl acetate. This way, the pH dropped harshly and the time in the acidic phase was shortened drastically. As seen in the lowest spectrum of Figure 12, this method was successful, showing only the set of peaks of the free acid and 2 singlets of the MOM-group.

Scheme 14: Final desymmetrization of the template, to obtain compound 27.

(21)

After the successful hydrolysis, the PFP-ester could be made. This was done according to a

procedure, frequently used in this research group. Compound 11 was obtained pure. As expected, the 19F-NMR showed three sets of peaks. This showed the nice asymmetric configuration of this compound, since both PFP-esters are in a slightly different environment, as seen in Figure 13 .

Now both, the ring fragment and the asymmetric template were obtained, these compounds could be coupled to form the desired chiral planar pre[2]rotaxane.

B Free acid

2 times 2 MOM-peaks

A

B

C

Figure 13: 19F-NMR of the syntesised compound 11 (PFP-ester). As expected, three sets of peaks were observed, due to the asymmetric structure. The doublet at A comes from the fluorine atoms ortho to the phenol, triplet B from the fluorine atoms meta to the phenol and triplet C from the fluorine atom para to the phenol.

Figure 12: 1H-NMR spectra of compound 10. As can be seen in the upper spectrum, three sets of aromatic peaks are observed. The second spectrum shows no product peak in the aromatic area, while the impurity peaks increased and a splitting of the mom-peaks. Spectrum 3 shows only product peaks and the absence of splitting in the MOM-peaks means that immediate extraction is the best procedure.

A B

A

D C

(22)

Synthesis of the pre[2]rotaxane

After finishing both the template as well as the ring fragment, now the pre[2]rotaxane could be synthesized. This was done in a two-step synthesis. At first this synthesis was done with the

symmetric template. This had two big advantages. It was already available on the lab and, because of symmetric structure of the template, the compound would be completely symmetric as well. This way, the 1H-NMR spectra would be much cleaner, since every similar proton would be in the same surrounding.

In the first step, 1 equivalent of template 30 and 2 equivalents of ring fragment 24 were solved, together with Cs2CO3 and 4 Å MS9,34. Under slightly heating, an esterification occurred. Despite the

fact that the PFP-ester is a relatively weak activated ester, the reaction was high-yielding.

With this compound 31 in hand, the most important reaction of this synthetic plan could be tested. The starting material was solved, together with TBTA, in dichloromethane. This was done in a 1mM scale to prevent the starting material from polymerising. After degassing, a Cu-complex was added. The cyclization of the ring would be achieved via a CuAAC click reaction35. The first proposed catalytic cycle was proposed by Liang36 and Fokin37, both showing the coordination of both the alkyne and the azide to the same copper-complex.

Scheme 17: Proposed catalytic cycle of the CuAAC Click reaction. As

proposed by Fokin38, Astruc39 and Leigh40, this cycle makes use of 2

copper complexes, to finally form the triazole.

Scheme 16: Synthesis of the symmetric pre[2]rotaxane, by using two equivalents of the asymmetric ring fragment and one equivalent of the symmetric template.

(23)

However, others proposed a cycle using two copper centers38,39,40. In their proposed mechanism, a second copper complex binds to the triple bond, before coordination of the azide to both the triple bond as the 2nd copper-complex. This can be seen in Scheme 17. TBTA was shown to be a powerful stabilizing ligand for Copper(I). It protects from oxidation and disproportionation. On the other hand, it enhances its catalytic activity.

This reaction turned out to be successful. A yield around 50% showed that the two ring fragments were close enough to each other, to form the desired product easily

enough. If the yield would be much higher, even a 26-membered ring could have been tried. With a much lower yield, it meant that both ring fragments were too far from each other. The resulted formation of 48% of pre-rotaxane 32 was confirmed by high temperature 1H-NMR and LCMS. Due to the conformal possibilities of the macrocycle, NMR experiments at room temperature gave inconclusive spectra. This showed elevated temperatures were necessary. At 120oC, the spectrum showed sharp peaks at the expected shifts. The peaks of the terminal alkyne were gone, while a new peak of the triazole-C-H appeared. The LCMS also showed 1 clear peak at 852.33 (m/z). However, at this moment, still 2 possible compounds could have been obtained. The first was desired compound 32. The second possible compound, was the synthesis of a bicyclic compound 32b, as seen in Figure 14. This could not be concluded from the NMR-experiments and the mass spectrum.

To determine the compound that was formed, the template had to be cleaved. This way, it could become clear via HRMS, if there was a large ring or two small

rings present. To obtain this, the template was cleaved by Tesser’s base (a mixture of dioxane : MeOH : 2M NaOH (3:1:0.1)), according to a procedure by Tesser himself9,41, previously used in this research group.

After this reaction, a 1H-NMR was measured, showing a mixture of the free ring and the template, containing more template, since this solved easier in chloroform, as can be seen in Figure 15. In this figure, spectrum of the crude mixture, it is showed that both the chemical shifts change as well as the sharpness of the peaks changes completely. As described before, the spectrum of compound 32 only showed broad inconclusive peaks at rt, while the spectrum in Figure 15 shows sharp peaks again. This fact already showed, cleavage of the template. TLC showed 2 clear spots as well. No third spot was seen, which was expected if both the large as the small ring were present. An LCMS

experiment was performed, showing only 1 peak at 611.27 (M+H). No peak was observed around 305.14, where the small ring was expected. After purification, the free large ring was obtained pure. This was confirmed by HRMS and 1H-NMR, since all peaks, belonging to the template were gone.

Figure 14: Possible bicyclic side-product

(24)

Now the synthesis of the symmetric pre-rotaxane was concluded, a new synthesis could be

performed to obtain the asymmetric compound. This should be done via the same procedure. Only this time, we expected to obtain a racemic mixture of both enantiomers.

In the first step, the coupling was as high-yielding as the previous time with the symmetric template. The first time, the second step was performed, it gave a yield of 50%. The second time however, it gave a much higher yield. The possible explanation to this unsuspected success was actually found quite fast. Starting this reaction, a calculated amount of DCM was used, for the reaction to have a 1mM solution. The reaction was started in the morning. At the end of the day, the amount of DCM was still the same. Somehow, during the night, the DCM started to slowly evaporate, which might have been caused by fluctuation of the nitrogen gas pressure. Due to this evaporation, the

decreasing amount of unreacted compound might have encountered the complex easier, to form the triazoles as well. Due to the already formed, large amount of product, the risk to obtain oligomers and polymers decreased as well, ending up with a higher yield, than normal wise obtained with a consistent 1mM solution. The analysis of this compound was done with 1H-NMR at 120oC and HRMS (see Figure 16) again. Unfortunately, the 1H-NMR still wasn’t conclusive.

Figure 15: 1H-NMR of the crude product. As described, the integrals of the free ring are normalised to their respective amount of protons. The peaks of the template give higher integrals, due to easier solution.

B

A C+D E F+G H I J K L

(25)

As seen in the HRMS spectrum, only 1 peak was found at 856.3314 (m/z). A small peak was found at half the mass due to the doubly charged molecular ion. As the cleaving of the template in the symmetric product clearly showed, only the large ring was formed, this assumption was made for this compound as well. This would also be checked in the next steps. During those steps, 1 of the protecting groups would be cleaved, followed by an esterification with the enantiopure R-configuration of the Mosher acid chloride. This way, the current racemic mixture, would be

converted from two enantiomers into a mixture of two diastereomers. If successful, this would also confirm the large ring, since the bicyclic compound 33b isn’t chiral. This way, there would be no diastereomers.

Next to the NMR and MS-experiments, other techniques were used to analyse this compound. The first technique was the use of a chiral HPLC. Because of the two formed enantiomers of compound

17, there might be a possibility to separate them on a chiral column to obtain more evidence for the

formation of the desired compound. This was firstly done on a small chiral HPLC at the UvA.

However, this gave no result. A sample of the compound was sent to Symeres (previously named as Mercachem). This company has a group, specialized in chiral separations. After using multiple

different normal phase (NP) columns, no retention was obtained. Due to solvation issues in alcohol, a method with DCM was tried. This showed some separation, but far from a good retention. To get this, reversed phase (RP) chiral columns were tried. This is because these columns are better to withstand other solvents than alcohol. On the Chiral-V column, a small separation was observed, as can be seen in Figure 17, using isocratic 98% acetonitrile. The two enantiomers came off at 1.62 and 1.97 at peaks of around the same height. This was confirmed by a method on the Chiral-CF column, using 90% acetonitrile. This method had a bad separation, but it showed two peaks at 1.8 and 1.9. Unfortunately, this retention wasn’t baseline separated, making it impossible to separate both enantiomers with prep-HPLC.

Figure 16: Mass spectrum of the desired compound. The large peak at 856.3314 was the M + H peak. At the doubly charged molecular ion mass (428.1532) only a small peak was observed as well. 1) Measured spectrum of the obtained product. 2) Calculated mass of the sample.

1

(26)

min 2 4 6 8 mAU 0 2 4 6 8

DAD1 A, Sig=270,100 Ref=off (N:\LCMS6\2020\12\14\1214_004.D)

1. 30 4 1. 38 4 1. 61 7 1. 96 8 2. 16 6 2. 36 9

The second technique was the use of X-ray analysis onto crystals. Therefore, crystals were grown. This was done by solving a small sample in DCM and putting it in a

surrounding of pentane. This way, the crystals slowly crashed out. This was sent to Dr. Martin Lutz of the Universiteit Utrecht. He did some experiments and ended up with a preliminary X-ray structure of the compound (Figure 18). However, a couple of preconditions had to be taken into account. Due to the two protection groups, and the ability of the triazoles to rotate, this gave a lot of disorder. This meant that in the preliminary picture, he made, this disorder wasn’t taken into account. Next to that, and to our surprise, the two protection groups are not sterically large enough to stop the template from rotating inside the large ring. In the model, that was made at the beginning of this project, the two protection groups seemed to be large enough to prevent this from happening.

However, this way, also the disorder of these two protecting groups was ignored as well. With these two major terms, the structure of the compound was made as can be seen in Figure 18. Here the dashed lines of the protection groups, are the minor (27%) disorder product and the solid lines show the major (73%) disorder product. Thanks to this picture, and the observation that the internal

template could rotate, it was also understandable that the separation on the chiral HPLC was harsh. Due to this rotation, there were no two clear enantiomers. While the rotation could be slow, even after a separation, the product could racemize again. To prevent this from happening, the addition of the already mentioned, sterically larger, Mosher ester was necessary.

Figure 18: Preliminary X-ray structure of the desired compound

(27)

Synthesis of the diastereomer with Mosher acid chloride

After the synthesis of the desired compound was confirmed, a next step was proposed. By cleaving one of the two protection groups of the inner template and addition of the Mosher acid chloride into an ester, diastereomers would be formed. This should also be seen in 19F-NMR, because there would be 2 peaks. Next to this, the rotation of the inner template would be prevented as well, due to the sterically larger group. To make this happen, a decision had to be made, which of the two protecting groups had to be cleaved. As seen before, the cleaving of the allyl-protection group gave quite low yields. However, it was tried on a 20mg scale with the procedure with phenylsilane. The better-working procedure couldn’t be used here, because the use of a base in methanol might cleave the esters as well. This gave a yield of 25%, which was low as expected.

Therefore, the deprotection of the MOM was tried, according to a procedure of Watanabe et al42.

By reacting compound 17 with concentrated HCl in methanol on a 20mg scale, 14mg of product was obtained. The 1H-NMR spectrum was just as difficult to understand as the spectrum of compound 17 itself. However, it was clearly seen that the two singlets, belonging to the mom-group were gone. To confirm, HRMS was performed. This showed indeed the synthesis of this compound with only one peak at M+H.

Now the synthesis of this compound was confirmed, the second step could be performed. The obtained deprotected compound was solved in DCM, together with triethylamine,

dimethylaminopyridine and the Mosher acid chloride according to a procedure of Nakajima et al43.

Scheme 20: Deprotection of the MOM-group from the inner template.

(28)

As Mosher acid chloride, the enantiopure (R)-configuration was used. This to prevent multiple peaks in the fluorine spectrum because of mixture of both configurations. By using only one of the two, only two peaks of both diastereomers were expected.

After work-up, 1H-NMR again was inconclusive and the amount of obtained product was too low to perform a 13C-NMR. Therefore all obtained material was solved in DMSO and a 19F-NMR was

measured. Expected was a spectrum with one set of two peaks corresponding to the diastereomers. However, by taking a spectrum at room temperature, 3 sets of peaks appeared. This was kind of confusing, but quite fast after this spectrum was obtained, a possible explanation was brought onto the table. As the X-ray structure already showed, there was a lot of disorder in the region of the triazoles. This meant that these could rotate. This problem could be the same in this compound. Both triazoles could rotate, which means that the side of the proton and the side of the two nitrogen atoms can both point towards one of the protection groups of the inner template. This means that the proton sides can both point to the allyl-goup, both to the Mosher-group and both to the other protection group. This of course would give three sets of peaks, since they would also still give the two diastereomers in each conformation.

To get a spectrum containing only one set of peaks, the temperature had to be increased. While doing this, a 1H-NMR was measured at 100oC, which made the spectrum a bit clearer. Despite the sharpness of the peaks in this spectrum, the integrals were still inconclusive. Therefore, at high temperature, only 19F-NMR experiments were performed (Figure 19). To get a better idea, at which temperature, only one stable conformation was obtained, the experiments were measured with 20oC differences from room temperature until 120oC. The results can be seen in Figure 19. In this figure, you can clearly see three sets of peaks at room temperature and at 40oC. At 60oC only on set of peaks is observed. At this temperature only one conformation is stable. At higher temperatures,

coalescence is observed.

Scheme 22: Three proposed sets of diastereomer conformations at room temperature. As shown, the triazole can rotate. Its conformation is almost perpendicular to the inner template

(29)

This already gave a good conformation that the desired product was obtained due to the fact that the peaks appeared in sets. This meant that both diastereomers were formed, what also showed that both enantiomers of the desired product were formed, since only one configuration of the Mosher acid chloride was used. As a last confirmation, also an HRMS experiment of this product was performed, showing a peak at exactly the right mass. There were no other peaks in the spectrum as expected.

(30)

Second route

For the second route, a more elaborate pathway was chosen. However, in the final structure, the same template would have been used as in the first route. To start with the ring fragment,

compound 37 is synthesized following a previously published procedure by Cho et al.44 and Tsubaki et

al.45,46. Due to the reaction of 4-tert-butylphenol with formaldehyde under basic conditions, the bis-hydroxymethylated compound 37 was obtained with a yield of 51%. Subsequently, the phenol and one of both benzylic alcohol groups where protected by 2,2-dimethoxypropane (DMP). This

acetonide compound 38 was obtained with a yield of 88%. This was necessary, to make sure, only the one free benzylic alcohol would react in the next step.

In the next steps, the benzylic alcohol had to be substituted for a terminal alkyne. Therefore, an adapted procedure, based on a procedure from Cho et al.44, was performed. To compound 38, one equivalent of both PPh3 and NCS were added in cooled THF. After stirring overnight, full conversion

was obtained with a yield of 78%. The amount of PPh3 had to be exact, since it’s presence would

interfere in the next step.

This step was the actual substitution with the terminal alkyne. For this reaction a Heck alkynylation (Copper-free Sonogashira) was employed47,48. In the original procedure, both the catalyst

(Pd(MeCN)2Cl2) and the ligand (XPhos), were added separately (as seen in the catalytic cycle (Scheme

25)), together with Cs2CO3 as base to a solution of compound 39 in THF. To this mixture, two

equivalents of TMS-actylene were added. This mixture was refluxed for 3 days. As seen in Scheme 25, in the first catalytic cycle Pd(II) is reduced to Pd(0) after which compound 39 binds via oxidative addition. Then, due to transmetalation with bis-alkynyl complex D4, formed in cycle 2, the chloride and the TMS-acetylene are substituted. The catalyst coordinates a new TMS-acetylene molecule to remake the catalyst. The product than cleaves from complex D2 via reductive eliminaton.

Scheme 23: Synthesis of compound 38.

(31)

However it was found, that synthesising the Pd(XPhos)2Cl2 catalyst first, before adding it to the

reaction, gave higher yields. This was done by solving both the catalyst as the ligand in dry THF, stir it overnight and evaporating the solvent. This way, the catalyst was obtained with a yield of 98%. In addition to this, the catalyst had to be made fresh before every reaction, to get the highest yields. During these reactions, It was also discovered that increasing the amount of TMS-acetylene to 3 equivalents would increase the yield. In the end, the most effective approach of this reaction ended with the desired product with a yield of 69%. However, the results from this reactions varied each time, which made this reaction quite inconsistent. This already gave some doubts about this synthesis route.

In the meantime, the desired tail fragment was synthesised. This was done in a two-step synthesis. In the first step, 6-bromohexanoic acid was solved in DMSO, according to a procedure from Pathak et

al.49. NaN3 was added to substitute the bromide for an azide. After a careful acidification, this

reaction was rather successful with a yield of 86%. In the second reaction, the acid was turned into a PFP-ester, via the same conditions as with the template. This reaction was successful as well, ending up with tail fragment 42 with a yield of 70%.

Scheme 25: Catalytic cycle of the Heck alkynylation (Copper-free Sonogashira), as reported by Gazvoda48.

(32)

After the synthesis of compound 40, the acetonide had to be cleaved again, to make the phenol and the second benzylic alcohol reactive again, whereafter the phenol had to be protected again. The preferred procedure for the first step was cleavage by hydrobromic acid in THF50. This reaction was successful. Unfortunately, the product was found to be too unstable to purify by column

chromatography the first time. The second time, according to 1H-NMR, compound 41 was obtained with a yield of 95% (not purified). This crude was used immediately in the next reaction.

This was the protection of the phenol, to ensure, no unwanted side reactions would occur in the next steps. This was done via the same reaction conditions as used for the allylation of the template. Despite, the full conversion of the starting material during this reaction, the desired product wasn’t obtained. The measured 1_{H-NMR showed no peak at all of the terminal alkene. However, something}

had happened, since the spectrum was clearly different from the starting material. Due to time shortage, the success of the first route and the earlier doubts about reproducibility, it was decided to leave this route as it was and put all effort in the first route.

Scheme 27: Proposed synthesis of 44. Unfortunately, after the successful synthesis of 43, no product of compound 44 was obtained at all.

(33)

Conclusion

During this project, the desired pre[2]rotaxane was obtained. Therefore two asymmetric compounds had to be made. These were obtained, with low yields, as expected.

The synthesis of the desired planar chiral pre[2]rotaxane was confirmed via X-ray analysis, mass spectrometry, chiral HPLC and the following steps to obtain diastereomers with the Mosher ester. Unfortunately, some drawbacks were found as well. The X-ray analysis showed the possibility for the template to rotate inside the large ring, due to the small steric hindrance of the protection groups. This was the opposite of what was expected after a model was built at the beginning of this project. The picture also showed major disorder of the triazoles, which was confirmed by the 19F-NMR of the Mosher ester, since, at low temperature, three sets of peaks were observed. Due to the rotation of the inner template, the separation over chiral HPLC gave no clear baseline retention. This means that, unfortunately, the product can’t be separated via prep HPLC for now.

Due to all of these results, some mixed answers could be given to the research question. The method, found during this project, is clearly a good way to obtain planar chiral pre[2]rotaxanes. As found, they could also be separated, if the inner template cannot rotate as it possibly can now. This means that by using this method, with more sterically hindered protection groups on the template, a good separation can be obtained. For the control of specific chirality in lassopeptides, this procedure is not as useful as we hoped at the start of this project. However, as a result of the obtained

knowledge and the possibility to do some adaptations to the obtained product, the found procedure can and will provide ways to also obtain the desired [3]rotaxanes, with specifically induced chirality.

(34)

Outlook

As described in the conclusion, some adaptations have to be made to the obtained product. Due to its possibility to rotate the inner template inside the large ring, the used protection groups have to be sterically larger. Therefore, as a first adaptation, one of the protecting groups needs to be

substituted by a stopper prior to enantiomeric resolution. The second adaptation to this procedure can also be the cleaving of the second protection group before sending the product to Symeres for separation on the chiral HPLC. As discussed with this company, it’s more likely to have good separation of the enantiomers, if one of the two phenols is unprotected. If the second one is protected by a stopper, this is no problem at all.

Another possible problem is the addition of the stopper as well as the connecting part. After obtaining the desired product, a thought came up, that the triazoles can interfere in the next steps. The problem could be the fact that after the protection groups are cleaved, the stopper and the connecting part have to be connected via ether-connections. However, the triazole has the ability to connect with this ether group as well, forming a quaternary nitrogen cation. This is possible, due to the fact that it’s free electron pair doesn’t take part in the aromatic system. Therefore, ether connections can be problematic. At the same time, esters can’t be used as well, since the template itself is attached via esters. To obtain the rotaxane from the pre-rotaxane, these esters have to be cleaved.

Therefore a new plan was suggested. In this plan, the protection groups of the phenols would be changed from the beginning. One of the protecting groups can be changed into the stopper immediately, as also described above. The second group can then be changed into the connecting part already. This connecting part can be almost the same as the allyl-group, that was used now, but multiple carbon atoms longer (like a C9-chain). If this is done before the template is bound to the ring

fragment, these groups don’t have to be cleaved, which avoids this problem. Since the template is suspected to be perpendicular to the macrocycle, this also wouldn’t interfere with the cyclization. It is not sure, if this product could be separated by prep HPLC. If this is possible, the connecting parts can be connected by a Grubbs metathesis of the terminal alkenes, forming specifically chiral [3]rotaxanes in a slightly different way, using the same building blocks.

(35)

Acknowledgements

At first I would like to thank Prof. Jan van Maarseveen for giving me the opportunity to do this exciting project at his group and the valuable ideas, discussions and help during this project. Second,

Simone Pilon for his guidance on the lab, his smart ideas during this project and the fun talks on the

lab! Nick Westerveld for synthesizing large batches of starting material, insightful talks about the project and all the fun ‘substantive’ conversations and Bas de Jong for making the work on the lab more streamlined. Prof. Joost Reek for being my second examiner. I would also like to thank Bart van

Leeuwen and Jessica Evers for their previous work on both routes, allowing me to start with a little

advantage. Of course I would like to thank Dr. Martin Lutz from the Universiteit Utrecht for his time to do X-ray analysis on my crystals and his knowledge about this and Colinda van Tilburg BSc. from Symeres for her time to measure my sample on several of their chiral HPLC columns. Of course I would like to thank the SOC group and SusPhos as well for the fun times during my presence at the group. Lastly, I would like to thank Ichelle Biesterbos, my parents and my brother for your love and being there, when reactions failed or whenever I needed to complain about the research.

(36)

Experimental

Reactions were carried out under air, without additional measures such as drying, unless stated otherwise. Thin layer chromatography (TLC) was performed on Merck TLC plates (0.25 mm) pre-coated with silica gel 60 F254. Column chromatography was performed using SilaFlash P60 (40-63 µm) under compressed air flow. Starting materials and reagents were used as supplied by

commercial vendors. Dry solvents were obtained pure via MBraun SPS 5/7. NMR spectra were recorded with Bruker DRX-300, 400 and 500 MHz instruments. Chemical shifts (δ) are reported in ppm relative to residual undeuterated solvent peaks and described as follows: Chemical shift (multiplicity, number of H). Abbreviations used for the multiplicity are: s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet), dd (doublet of doublet), td (triplet of doublet). High-resolution mass spectra (HRMS) were recorded on an AccuTOF GC v 4g, JMS-T100GCV Mass spectrometer (JEOL, Japan) and HR-ToF Bruker Daltonik GmbH (Bremen, Germany) using field desorption (FD) as ionization method.

Compound 18

This compound was already available on the lab, Therefore, it wasn’t synthesized for this research specifically.

Compound 19

To 100 mL DMF, 10g (54.5 mmol, 1 equiv.) 18 and 22.6g (163.5 mmol, 3 equiv.) K2CO3 were added. To

the solution 4.7 mL (6.6g, 54.5 mmol, 1 equiv.) allyl bromide in 15 mL DMF was added dropwise by syringe. The solution was stirred at rt. overnight. The solution was partitioned in EtOAc and H2O. The

aqueous layer was extracted with EtOAc (2x). The organic layer was washed with aq. sat. NH4Cl (2x),

H2O (2x) and brine (1x), dried over MgSO4, filtered and concentrated in vacuo. In the end 10.63g

(47.4 mmol, 87%) of a white solid was obtained. 1H-NMR (400MHz, CDCl3): δ 6.90 (s, 2H), 6.11 (m,

1H), 5.47 (dd, 1H), 5.30 (dd, 1H), 4.73 (S, 4H), 4.42 (d, 2H), 3.83 (s, 3H). 13C-NMR (400MHz, CDCl3): δ

156.23, 148.27, 135.14, 133.57, 117.94, 113.53, 75.70, 61.08, 55.62. IR (cm-1): 3264, 2994, 2873, 1564. HR-MS: calcd for C12H16O4 (M+): 224.1043, found: 224.1049

(37)

Compound 21

To 30mL dry DCM under N2-atmosphere at -10oC, 500mg (2.2 mmol, 1 equiv.) 19 was added. To the

solution 744 µL (540mg, 5.3 mmol, 2.4 equiv.) triethylamine in 5 mL DCM and 379 µL (560mg, 4.9 mmol, 2.2 equiv.) methylsulfonyl chloride in 5 mL DCM were added dropwise by syringe.

The solution was stirred for 2h. The reaction was quenched with H2O. The aqueous layer was

extracted with DCM (2x). The combined organic layer was washed with H2O (2x) and brine (1x), dried

over MgSO4 and filtered. DCM was evaporated in vacuo. In the end 819mg (2.1 mmol, 97%) of a

yellow oil was obtained. 1H-NMR (400MHz, CDCl3): δ 7.05 (s, 2H), 6.12 (m, 1H), 5.48 (dd, 1H), 5.35

(dd, 1H), 5.29 (s, 4H), 4.42 (d, 2H), 3.84 (s, 3H), 3.05 (s, 6H). 13C-NMR (400MHz, CDCl3): δ 156.15,

149.71, 132.78, 128.53, 118.63, 117.24, 66.31, 55.82, 38.06. IR (cm-1): 3026, 2939, 2845, 1600, 1333, 1099. HR-MS: calcd for C14H20O8S2 (M+): 380.0594, found: 380.0592

This compound was already available on the lab. Therefore, it wasn’t synthesized for this research specifically.

Compound 23

To 80 mL of dry THF under N2-atm., 3g (7.9 mmol, 1 eq.) compound 21 was added. To 60 mL dry THF

at 0oC under N2-atm. 694mg (7.97 mmol, 1 equiv.) azido-alcohol and 316,1mg (190mg, 7.9 mmol, 1

equiv.) 60% NaH were added. After stirring at 0oC for 15 min, the cooled alkoxide solution was added by addition funnel. After addition, the reaction mixture was stirred overnight at rt. The reaction was quenched with aq. sat. NH4Cl. The aqueous layer was extracted with DCM (2x). The combined organic

layer was washed with aq. sat. NH4Cl (2x), H2O (2x) and brine (1x), dried over MgSO4, filtered and

concentrated in vacuo. The concentrated azide was solved in 60 mL of dry THF under N2-atm. To 80

mL dry THF, under N2-atm., 629mg (377mg, 15.7 mmol, 2 equiv.) 60% NaH and 1.3mL (1.2g, 17.2

mmol, 2.2 equiv.) were solved. The mixture was stirred at 0oC for 30 minutes. After stirring, the cooled solution was added by addition funnel. The mixture was stirred overnight at rt. The reaction was quenched with aq. sat. NH4Cl. The aqueous layer was extracted with DCM (2x). The combined

organic layer was washed with H2O (2x) and brine (1x), dried over MgSO4 and concentrated in vacuo.

A column was performed using PE:EtOAc 10:1 -> 5:1. In the end 500mg (1.45 mmol, 18%) of a clear colourless oil was obtained. 1H-NMR (400MHz, CDCl3): δ 6.96 (s, 2H), 6.10 (m, 1H), 5.44 (dd, 1H), 5.25

(dd, 1H), 4.62 (s, 2H), 4.60 (s, 2H), 4.36 (d, 2H), 3.82 (s, 3H), 3.71 (t, 2H), 3.66 (t, 2H), 3.44 (t, 2H), 2.53 (td, 2H), 2.00 (t, 1H). 13C-NMR (400MHz, CDCl3): δ 156.08, 148.71, 133.83, 132.26, 132,03, 117.25,

114.56, 113.79, 81.35, 75.96, 69.36, 69.24, 68.49, 68.07, 67.87, 55.59, 50.89, 19.90. IR (cm-1): 3293, 2865, 2102. HR-MS: calcd for C18H23N3O4 (M+): 345.1783, found: 345.1689

Template assisted synthesis of a planar chiral pre[2]rotaxane