Chapter 1: Macrocycles, Catenanes and Rotaxanes

Enzymes, nature's catalysts, are remarkable. Operating under mild conditions, each molecule of enzyme is able to process a large number of substrate molecules. They typically display high levels of selectivity both in recognition of substrate being bound and in the regio- and stereochemical outcome of the chemical reaction catalysed. Scientists have therefore sought to establish and explain how enzymes achieve these feats. For example, the origin of substrate selectivity was famously postulated by Fischer as being comparable to a “lock and key”.¹  It is now recognized that both enzyme and substrate undergo significant conformational changes upon binding of the substrate within the active site of the enzyme, as described in Koshland's induced fit theory.² 

Chemists from a multitude of chemical sub-disciplines have aimed to replicate and build upon the feats of natural enzymes by preparing synthetic catalysts – both as biomimetics or purely for synthetic chemical purposes. A key challenge in emulating the success of enzymes with synthetic catalysts is to achieve the necessary arrangement of functional groups required for binding and reaction.

One solution is to deploy macrocycles (Figure 1.1A).³  Macrocycles possess cavities where binding may occur, and many examples have appropriate “functional handles” to which reactive functionality may be incorporated. More recently, interest has turned towards using interlocked molecules as reactive partners. By possessing macrocycles within their interlocked structures, catenanes⁴  (Figure 1.1B) and rotaxanes⁵  (Figure 1.1C) are natural descendants of macrocycles. By being interlocked, they are inherently three-dimensional, and there is the possibility of large-amplitude motion of their constituent components as well as chirality arising from the mechanical bond.

This chapter surveys examples of reactivity arising in the confined spaces of macrocycles, catenanes and rotaxanes. It is not an exhaustive account of the area, and the focus of the chapter is on the exciting recent developments in the application of catenanes and rotaxanes, which reflects the author's own research interests. Overall, the author is principally aiming to inspire the reader to investigate the literature further and perhaps, through their own work in the laboratory, add to the excellent chemistry to be found in this chapter.

Macrocycles featured heavily in the work of the 1987 Chemistry Nobel Laureates Pedersen,⁶  Cram⁷  and Lehn.⁸  Amongst the work from the laboratories of Cram and Lehn are early examples of utilizing derivatives of macrocyclic crown ethers as reactive sites. More “container”-like macrocyclic frameworks such as cyclodextrins,⁹  calixarenes¹⁰  and cucurbiturils¹¹  have also been used for this purpose.

As already highlighted, this chapter cannot be exhaustive, and this is particularly the case for examples of chemical reactivity within the confined spaces of macrocycles, which is a very large body of research. This section therefore only serves as a selective illustration. Specific examples from the primary literature are discussed, with references to recent reviews also being provided where relevant.

In 1979, a transacylase mimic based on a crown ether was reported by Cram and co-workers (Figure 1.2).¹²  The crown ether 1 can bind protonated ammonium (–NH₃⁺) groups via geometrically complementary hydrogen bonds to a pseudo-18-crown-6 macrocycle. The axially chiral dinaphthyl leads to enantioselective recognition of amino ester substrates through steric interactions, while the presence of thiol groups leads to transfer of the acyl group from the substrate. For (S)-1, host/l-amino acid ester pairs are stabilized in comparison to host/d-amino acid ester pairs. The degree of selectivity, as measured by the rate of reaction for (S)-1 versus (R)-1 depends on the steric bulk of the α-substituent of the l-amino acid p-nitrobenzoate ester. For example, for R = Me (alanine), there is no observable selectivity (k_S/k_R = 1), while for R = CHMe₂ (valine), k_S/k_R = 9.2. While this chiral recognition is impressive, it should be noted that hydrolysis of the acylated host product proves to be very slow, and so this system does not exhibit catalytic turnover.

The group of Lehn subsequently reported on the use of aza-crown ether [24]N₆O₂ 2 as a model system for the hydrolytic enzyme ATPase (Figure 1.3).¹³  When protonated, macrocycle 2 can bind anions in aqueous solution through electrostatically augmented hydrogen bonds. In particular, the macrocycle can bind adenosine triphosphate (ATP) and accelerate its rate of hydrolysis over a wide pH range, as monitored by ³¹P NMR spectroscopy. The proposed mechanism for the hydrolysis of ATP by macrocycle 2 begins with deprotonation of one ⁺NH₂ site of 2.6H⁺, to allow for nucleophilic attack at the terminal P atom of bound ATP. This leads to phosphorylation of the macrocycle with the resulting adenosine diphosphate (ADP) being bound. In this case, catalysis is possible due to good turnover arising from the greater affinity of 2 for ATP compared to ADP and hydrolysis of the phosphorylated intermediate macrocycle. In a later report, Hosseini and Lehn demonstrated that macrocycle 2 can catalyse the reverse process, that is, convert ADP to ATP, by using an excess of acetyl phosphate and Mg(ii) as a promoter, and as such [24]N₆O₂ 2 can also be viewed as a potential ATP synthase mimic.¹⁴ 

Cyclodextrins (CDs) have been widely studied as functional macrocycles due, in part, to their ready availability. They are water-soluble, cyclic oligomers of α-d-glycopyranoside monomers (α-CD consisting of 6 monomers, β-CD 7, γ-CD 8), and possess a hydrophobic inner cavity and two hydrophilic outer rims decorated with hydroxyl groups. Chemists have studied both underivatized and functionalized cyclodextrins as potential catalysts.¹⁵ 

A seminal example from Breslow and Campbell involved the chlorination of anisole by use of hypochlorous acid (Figure 1.4A).¹⁶  In the absence of α-CD, anisole is chlorinated at both para and ortho positions (in a ratio of 1.48 : 1). However, in the presence of an excess of α-CD, para substitution is hugely favoured (ratio of 21.6 : 1 in the presence of ∼100 equivalents of α-CD when 72% of anisole is bound). The enhancement of selectivity is explained by the authors as arising from the binding of the anisole molecule within the cavity of α-CD, shielding the ortho positions.

Functionality may be added to either the wide or narrow rim of a CD. An early example of where functionality was added to the wider rim of α-CD involves attachment of a Ni(ii) complex, compound 3 (Figure 1.4B).¹⁷  The p-nitrophenyl acetate substrate is bound within the CD cavity of 3 and is hydrolysed by interaction with the Ni(ii) centre and CD hydroxyl groups, for example, as illustrated, by the attack of the oxime ligand.

A more recent example of narrow rim functionalization is depicted in Figure 1.5.¹⁸  Here a larger β-CD has been substituted with pyridoxamine to generate 4. While pyridoxamine can transform α-keto acids to α-amino acids by itself, by being attached to the β-CD, transamination of aryl substrates such as phenylpyruvic acid occurs much more rapidly than for pyruvic acid, due to the greater affinity of the β-CD cavity for aryl substrates. Furthermore, due to the chirality of the β-CD, a significant degree of chiral induction is possible. For example, in the formation of phenylalanine, an enantiomeric excess (ee) of 52% was observed.

Calix[n]arenes are cyclic oligomers consisting of n 2-methylene-1-phenol units. Like cyclodextrins, calixarenes typically possess two different-sized rims – an upper hydrophobic aromatic rim and a lower hydrophilic rim of phenol groups. Both rims may be functionalized to allow for their use in catalysing chemical reactions.¹⁹ 

For example, Reinhoudt and co-workers demonstrated efficient phosphate diester transesterification by the calix[4]arene-based dinuclear Zn(ii) catalyst 5 (Figure 1.6A).²⁰  The two Zn(ii) sites are placed on the upper rim of the calixarene and can facilitate a 23 000-fold rate enhancement in the cyclization of 2-(hydroxypropyl)-p-nitrophenyl phosphate (HPNP), which may be considered to be an RNA model substrate. The researchers ran control experiments with a monofunctionalized calix[4]arene analogue and a simple reference Zn(ii) pyridine complex in order to confirm the contributions of the calix[4]arene and the two Zn(ii) centres in the function of 5 with respect to substrate binding and catalysis.

While chiral induction in calixarene-mediated reactions is possible by appending chiral groups to lower rim hydroxyl groups,²¹  a recent line of research interest has been the study of inherently chiral calixarenes.²²  Shimizu and co-workers prepared inherently chiral calix[4]arene 6, which, once resolved into optically pure enantiomers, was used as an organocatalyst in asymmetric Michael addition reactions of thiophenols (Figure 1.6B).²³  With 1 mol% of enantiopure calix[4]arene 6, such reactions had a maximum observed ee of 31% (with almost quantitative yields of conjugate addition product).

Cucurbiturils, or CB[n]s, consist of n glycoluril units connected by two methylene bridges and are formed by the copolymerization of formaldehyde, glyoxal and urea. As better methods of preparing different-sized CBs have become more readily established, they are now being increasingly studied as catalysts.²⁴  A famous example was the use of CB[6] in the mediation of alkyne/azide cycloaddition reactions (Figure 1.7),²⁵  which predates the establishment of the famous Cu-catalysed alkyne/azide cycloaddition (CuAAC) “click” reaction.²⁶  In Mock's seminal report, alkyne and azide components are substituted with ammonium ions, which form strong ion–dipole interactions with the carbonyls at the rims of the CB[6] molecule. The alkyne and azide are then held in close proximity within the cavity of CB[6], leading to a rate acceleration for cycloaddition of 5.5 × 10⁴.

The encapsulation of reactants within CBs in a range of photochemical reactions has also been studied. To illustrate, E-diaminostilbene may be dimerized in CB[8] – with good selectivity for the syn-adduct (Figure 1.8).²⁷  The product could be released from the cavity of CB[8] by adding NaOH, allowing for recycling of the macrocycle. In the absence of CB[8], the main product was simply Z-diaminostilbene, the geometric isomer of the starting material.

The derivatization of cucurbiturils is still rather limited in comparison to cyclodextrins and calixarenes. Thus achieving asymmetric reaction outcomes using cucurbiturils might appear unlikely as the standard parent CBs are achiral. However, Scherman, Herrmann and co-workers have been able to report upon using CB[8] as part of a “supramolecular nanoreactor” catalysing asymmetric Lewis acid-promoted Diels–Alder reactions (Figure 1.9).²⁸  Using the chiral amino acid abrine (a methylated derivative of tryptophan), Cu(ii) as a Lewis acid, and CB[8] as encapsulating host, an ee of up to 92% was observed.

While statistical²⁹  and directed covalent³⁰  methods were used to synthesize interlocked molecules during the 1960s and 1970s, the field only truly began to flourish after Sauvage's 1983 report of the use of a Cu(i) cation to template the formation of catenanes.³¹  Sauvage's team prepared catenane 9 by self-assembling phenanthroline thread 7 through phenanthroline macrocycle 8, before cyclizing the second ring by Williamson ether synthesis (Figure 1.10). The Cu(i) cationic template could then be removed by treatment with cyanide.

Sauvage's use of a Cu(i) cation is now viewed as an example of passive metal templation – the role of the cation is simply to arrange the necessary precursors in space – and the covalent capture event that forms the interlocked molecule is happening outside the macrocyclic cavity. Leigh and co-workers have since pioneered active metal templation.³²  Here the cation is not only bringing together components but also acting as a reagent or, better, a catalyst for the covalent bond-forming reaction that leads to the formation of the interlocked molecule. The most commonly used reaction for active metal templation is the CuAAC “click” reaction, as illustrated by the Leigh group's seminal rotaxane synthesis in Figure 1.11A.³³  The Cu cation can bind to the pyridyl N of the macrocycle and simultaneously coordinate to both axle precursors and then catalyse the dipolar cycloaddition reaction within the macrocycle cavity. Goldup's team further optimized this reaction, achieving excellent yields of rotaxane 16 by use of a small bipyridyl macrocycle 13 while using only stoichiometric equivalents of pre-axle components (Figure 1.11B).³⁴  Catenanes may also be prepared by active metal templation, as first demonstrated by Leigh's group in 2009.³⁵ 

Reacting components within a macrocyclic cavity to “snap” shut an axle to form a rotaxane is not limited to active metal templation. Indeed, predating Leigh's active metal template work, Vögtle and co-workers reported on the synthesis of rotaxane 20 (Figure 1.12).³⁶  Phenolate 18 may be held by hydrogen bonding within the cavity of isophthalamide macrocycle 17 and then react, by simple nucleophilic substitution, with the dibenzyl bromide 19 to afford 20 in excellent yield.

Very recently, Leigh's group has reported on the preparation of rotaxanes by hydrogen bond-mediated transition state stabilization. For example, cyclic sulfate and amine half-axle components 22 and 23 were reacted in the presence of pyridyl-2,6-dicarboxamide macrocycle 21 (Figure 1.13).³⁷  With a five-fold excess of axle components, 70% yield of rotaxane 24 was produced. This is despite neither half-axle component associating significantly with the macrocycle in the absence of the other half-axle component. It has subsequently been demonstrated that analogous reactions may be run using commercially available 24-crown-8 macrocycle 25 (Figure 1.14).³⁸  The crown ether is able to stabilize the transition state when a primary amine (26) reacts with a variety of electrophilic partners (e.g. 27). Even when only using stoichiometric amounts of half-axle components, yields of >50% are possible.

There have been several occasions where the presence of the macrocyclic component has affected the reactivity of the axle of a rotaxane, leading to the production of new interlocked molecules. Here two recent examples are presented to show how the reactivity of one interlocked component (the axle in both cases) is mediated as being part of a rotaxane.

Berna and co-workers serendipitously discovered that an interlocked N-benzylfumaramide (e.g. 29) will cyclize to form a β-lactam (30) in high yields by use of Cs(i) bases in DMF (Figure 1.15A).³⁹  The cyclization proves to be highly regio- and diastereoselective – in contrast to that of the non-interlocked fumaramide axle component. Recently, Goldup and co-workers have systematically investigated a rearrangement that occurred during what was anticipated to be a standard active template CuAAC rotaxane synthesis – which resulted in the production of rotaxane 33 containing an acrylamide axle (Figure 1.15B).⁴⁰  Considering the acrylamide was not observed in the non-interlocked axle by-product and having demonstrated that the acrylamide rotaxane is not formed by a reaction from the targeted triazole rotaxane itself, the authors proposed that the interlocked architecture stabilizes the Cu(i) triazolide intermediate of the CuAAC reaction and diverts the reaction down a rearrangement pathway to acrylamide rotaxane 33.

Due to their unusual interlocked molecular architectures, catenanes and rotaxanes present a number of structural features that may be used to achieve sophistication in catalysis. With their innate three-dimensionality, cavities are available within or between the interlocked components that could bind substrates and/or stabilize transition states.⁴¹  As celebrated by the 2016 Nobel Prize in Chemistry, there is also the possibility of motion of the interlocked components.⁴²  This motion can be relatively subtle, but several groups have looked to deploy stimulus-controlled large amplitude motions to allow for switchable (e.g. on/off) or processive catalytic behaviour. Furthermore, some groups are beginning to exploit chirality arising from the mechanical bond to achieve stereocontrol in product formation.⁴³ 

In this section, a wide selection of examples from the literature are discussed, categorized by degree of motion exhibited in their operation as catalysts or reactive scaffolds – whether (predominantly) static, switching or processive.

In 2004, Takata et al. reported on rotaxanes possessing axially chiral binaphthyls to create chiral environments for asymmetric benzoin condensations (Figure 1.16A).⁴⁴  Good yields (of up to 90%) were observed in the condensation of benzaldehyde (35) but with only modest enantiomeric excesses (< 32% ee) with rotaxane (R)-34. However, it should be noted that the chiral group for this rotaxane is part of the macrocycle; thus the chiral presence is being transmitted between the interlocked components to the catalytic thioazolium site on the axle. More recently, Takata reported on related rotaxane catalyst 37 for the O-acylative asymmetric desymmetrization of meso-1,2-diols (Figure 1.16B).⁴⁵  Very impressively, selective monobenzoylation of meso-1,2-hydrobenzoin 38 was achieved in quantitative yield and 98% ee.

Osakada and co-workers reported on [3]rotaxane 40 (the [3] prefix refers to 40 consisting of three interlocked components) for ring-closing Mizoroki–Heck reactions (Figure 1.17).⁴⁶  Each macrocycle of the rotaxane was appended with a Pd(ii) complex that could catalyse the Mizoroki–Heck reaction. The researchers specifically studied a cyclization process and found that the rotaxane gave higher yields of the targeted cyclized product 43 than when they used the equivalent quantity of Pd(OAc)₂. They proposed that by using rotaxane 40, the cyclization event is favoured over the formation of oligomeric by-products due to the reactants and intermediates being suitably aligned by the pre-organized Pd centres.

Loeb and Stephan demonstrated that a sterically unencumbered aniline can become a bulky Lewis base by converting it to a rotaxane.⁴⁷  The Lewis base donor, surrounded by a protective macrocyclic ring – either a 24-crown-6 as in 44 (Figure 1.18), or a 22-crown-6 – exhibits the reactivity of a frustrated Lewis pair, as exemplified by the activation of hydrogen gas in the presence of B(C₆F₅)₃, where the amine is protonated and the hydridoborate anion [HB(C₆F₅)₃]⁻ is formed.

Rotaxane (R, R)-45, prepared by Goldberg active metal template synthesis in Leigh's laboratory, has been deployed in enantioselective nickel-catalysed conjugate addition reactions (Figure 1.19).⁴⁸  Compared to an acyclic ligand analogue, rotaxane (R, R)-45 catalyses the reaction of ethyl acetoacetate 46 and trans-β-nitrostyrene 47 with a much better enantiomeric ratio (93 : 7 compared to 68 : 32) but considerably slower reaction time (27 versus 2 days for full conversion). In other words, the rotaxane improves the expression of chirality arising from the two stereogenic carbon atoms (the interlocked structure reduces degrees of freedom) but also restricts access to the cation (which is buried within the interlocked structure).

Leigh's group has also reported on chiral rotaxane catalyst (S)-49 (Figure 1.20).⁴⁹  Rotaxane (S)-49 is chiral due to the macrocycle being restricted to one end of the axle (so-called point mechanical chirality). The secondary amine “blocking group” at the centre of the axle may enantioselectively catalyse Michael addition and enamine reactions. Rather modest enantiomeric ratios were reported – but this can be explained in part by the rotaxane only being prepared in 84% ee.

Recently, an interesting result has been reported by Berna and co-workers (Figure 1.21).⁵⁰  Fumaramides bearing l-prolinamide fragments may catalyse asymmetric addition of ketones to β-nitrostyrenes. Remarkably, the enantioselective course of these reactions may be reversed by incorporation of the fumaramide as an axle of a rotaxane (e.g. 53). An accompanying DFT study provided insight into this observation. The calculations revealed that in the absence of the macrocyclic ring, the fumaramide NH group of the axle is able to participate as a hydrogen bond donor to the reaction substrate, and hence the lowest energy reaction pathway for the organocatalytic process is altered such that the other enantiomer of product is favoured.

Niemeyer and co-workers prepared the enantiopure binaphthyl-phosphoric acid catenane (S,S)-56 that is chiral by virtue of the chiral axis of the binaphthyl (Figure 1.22).⁵¹  The researchers demonstrated its use in asymmetric transfer hydrogenation reactions and found that catenane (S,S)-56 had much improved stereoselectivity when compared to the non-interlocked macrocycle, with comparable yields (but notably slower reaction times). A DFT study provided computational evidence that the high stereoselectivities for catenane (S,S)-56 were due to the interlocked structure.

Niemeyer has subsequently reported on a set of rotaxanes also containing a binaphthyl-phosphoric acid-bearing macrocycle and a secondary ammonium axle (Figure 1.23).⁵²  These heterobifunctional rotaxanes were studied as catalysts for asymmetric Michael additions to α,β-unsaturated aldehydes. In the first generation, rotaxane (S)-59 proved to be a more active catalyst than a combination of the non-interlocked macrocycle and axle; however, stereoselectivities were low. DFT calculations indicated that inclusion of bulky substituents on the rotaxane structure would enhance stereoselectivity; bulky isopropyl groups were added, and a dramatic increase in ee was observed for the resulting rotaxane (S)-60.

As detailed over two papers, Goldup and co-workers have attempted to make use of the three-dimensionality of rotaxane structures to sterically influence the outcomes of Au(i)-catalysed reactions, which is often difficult to achieve with simpler ligands due to the linear coordination preference of Au(i).^53,54  In their first paper, they focused on an achiral rotaxane 63, which incorporated the Au atom at the terminus of its axle (Figure 1.24A).⁵³  Unexpectedly, the rotaxane did not catalyse Toste–Ohe–Uemura cyclopropanation reactions even in the presence of the standard AgSbF₆ additive (used to abstract the chloride anion from the Au(i) cation) – unlike the non-interlocked axle. ¹H NMR spectroscopic evidence suggests that, upon chloride anion abstraction, the Au(i) centre interacts with the bipyridyl N atoms of the macrocycle. The researchers, therefore, added metal cations such as Zn(ii) or Cu(i) to disrupt this interaction, which did indeed lead to the switching “on” of the anticipated catalytic activity. Very recently this work has been taken further, by preparing an enantiopure chiral rotaxane (R)-64 (Figure 1.24B).⁵⁴  This rotaxane is chiral due to the incorporation of a rotationally asymmetric macrocyclic component onto the directional axle (mechanically planar chirality). The optimized conditions for the rotaxane to act as a catalyst in Toste–Ohe–Uemura cyclopropanation reactions are presented in Figure 1.24C. In an accompanying substrate screen, an ee as high as 74% was observed.

As already highlighted, appropriately designed interlocked molecules may possess large-amplitude motion of their components, controlled by a chemical or physical stimulus. Significant attention has been paid in the last decade to exploit this in catalysis.

In an interesting demonstration, Berná and co-workers prepared a rotaxane 68 that can catalyse Mitsunobu reactions (Figure 1.25).⁵⁵  As synthesized, the macrocycle predominantly resides over the azodicarboxamide functional group which is the best hydrogen bond acceptor site on the axle. Triphenylphosphine may attack the NN double bond, leading to the macrocycle shuttling along the axle and commencement of the Mitsunobu reaction process. Once the Mitsunobu reaction is complete, the rotaxane possesses a hydrazodicarboxamide group (69), which may be oxidized by iodosobenzene diacetate back to the original azodicarboxamide rotaxane 68, which may once again participate in the Mitsunobu reaction.

Leigh and co-workers have worked extensively on switchable rotaxane catalysts. For example, a series of papers have been published on the study and development of rotaxane 70, first reported in 2012 (Figure 1.26A).⁵⁶  As synthesized, the secondary amine is protonated, and the crown ether macrocycle resides at this site due to electrostatically augmented hydrogen bonding. However, deprotonation leads to translocation of the macrocycle, specifically shuttling between the equivalent triazoliums. The amine may then catalyse Michael additions (see Figure 1.26B) – even though a wide range of reactions may be catalysed by this functional group as disclosed subsequently.⁵⁷  Reprotonation leads to the macrocycle returning to the centre of the axle and the catalytic function switching off. However, it should be highlighted that when the crown ether is located on the ammonium station, the triazolium sites can participate in anion-binding catalysis.⁵⁸  Finally, by the inclusion of a stereogenic carbon atom next to the amine, Michael additions can be catalysed with enantiomeric ratios as high as 94 : 6.⁵⁹ 

Switching molecular machines by the use of light is attractive due to the potential to eliminate waste that is produced by repeated application of a chemical stimulus such as acids and bases. With this in mind, Berna and co-workers prepared rotaxane 74 whose catalytic activity is modulated by light (Figure 1.27).⁶⁰  With the macrocycle residing at the fumaramide station, rotaxane 74 may catalyse a chalcogeno-Baylis–Hillman reaction, with a diastereoselective ratio of 80 : 20 (E/Z). Photoisomerization to the maleamide leads to translocation of the macrocycle to the thiodiglycolamide of the axle, switching off diastereoselectivity in the reaction.

Amongst the Leigh's group more recent work is another example that exploits light switching. Rotaxane 78 can be switched between two sites; in one direction by light, the other by catalytic acid (Figure 1.28).⁶¹  At either site, the rotaxane can catalyse the conjugate addition of aldehydes to vinyl sulfones. However, the position of the macrocycle determines whether the R or S enantiomer of the product is enhanced. In addition, rotaxane 78 affords greater enantioselectivities (up to 40% ee) in comparison to the non-interlocked axle (2–14% ee).

The vast majority of switchable rotaxane catalysts rely on shuttling of the macrocyclic component along the axle. Cheng, Chiu and co-workers reported on rotaxane 82 whose catalytic activity may be modulated by the addition and removal of Na⁺ cations, which induces pirouetting of the macrocycle around the axle (Figure 1.29).⁶²  At least three on/off cycles of the Michael reaction of ethyl acetoacetate and β-nitrostyrene can be performed. In the “on” state (in the presence of sodium cations), the authors proposed that the enolate coordinates to the bound sodium cation, positioning it in proximity to the β-nitrostyrene, which is activated by coordination to a protonated tertiary amine of the macrocycle.

To complete this section of the chapter, attention is turned to two very recent reports on remarkably similar rotaxanes that are being deployed in two quite different catalytic processes. First, in a collaboration between the laboratories of Beer and Williams, rotaxanes 83–85 were investigated as catalysts for lactide ring-opening polymerization (ROP) (Figure 1.30).⁶³  As synthesized, the protonated rotaxanes are catalytically inactive. Once deprotonated, the amine N lone pair can coordinate to the alcoholic proton of the benzyl alcohol that acts as the initiator for ROP. The catalytic activity of the rotaxane depends on the second site on the rotaxane axle. When this is a thiourea (83), the crown ether slowly shuttles between the two-axle sites, the amine is less basic, and ROP is slow. Using a triazole as the second site (84) results in much faster macrocycle shuttling, and so ROP is fast. The same effect can be alternatively achieved by incorporating a bulkier stopper next to a thiourea (85). Only rotaxane 83 shows high isoselectivity in polylactide formation – this is tentatively attributed to the combination of the stronger binding of lactide by the urea of this rotaxane and the slow shuttling, allowing for the orientation of the growing polymer chain.

Almost contemporaneously, Leigh et al. reported on a switchable rotaxane 88 whose catalytic activity may be controlled by the addition of a chemical “fuel” (Figure 1.31A).⁶⁴  The “fuel” is trichloroacetic acid which is itself decomposed by the rotaxane to form chloroform and carbon dioxide. Adjusting the level of fuel added controls how long the rotaxane remains protonated. While protonated, rotaxane 88 may catalyse a hydrogen transfer reaction (Figure 1.31B), which was studied over single fuel pulses (of varying length) and multiple fuel cycles.

In 2003, Rowan and Nolte reported on [3]rotaxane 90, consisting of two Mn-metallated porphyrin containing macrocycles that were able to shuttle up and down a polybutadiene polymer axle, oxidizing alkene moieties as they did so (Figure 1.32).⁶⁵  In the presence of an oxygen donor, a Mn(v)O species forms, which can transfer the O atom to the alkene substrates in the central portion of the axle. The presence of an axial tert-butyl pyridine ligand on the metallated porphyrin is required to direct oxidation to occur within the macrocyclic cage cavity. At the time of the original report, it was not clear how the macrocycle moved and reacted with the axle, but in a subsequent study, evidence indicated the reaction proceeds via a random sliding process.⁶⁶  Subsequently, the group refined the design of the macrocyclic component, incorporating urea-terminated tails that efficiently shield the manganese porphyrin and direct the oxidation process to the inside of the catalyst cage, without need for the additional tert-butyl pyridine ligand.⁶⁷ 

To conclude, Leigh's group prepared rotaxane 91 as a synthetic analogue of a peptide synthesizer (Figure 1.33).⁶⁸  The macrocyclic ring possesses a thiolate group that iteratively removes amino acids, in order, from the axle and transfers them to a peptide-elongation site through native chemical ligation. The sequence of amino acids – as set by the axle – is maintained in the peptide chain generated, which may be liberated from the macrocycle component by simple hydrolysis at the end of the reaction. A more efficient synthesis where a simple [2]rotaxane was prepared first, before adding an axle extension with three amino acids installed, was subsequently reported.⁶⁹  Impressively, the incorporation of non-proteinogenic β-amino acids into a growing peptide chain has also been demonstrated.⁷⁰ 

The cavities of macrocycles and macrocycle-containing interlocked molecules continue to be investigated as sites for chemical reactions over half a century since the first reported examples using cyclodextrins. While more challenging to prepare, catenanes and rotaxanes offer interesting opportunities with their three-dimensional structures, including chirality arising from the mechanical bond and large amplitude molecular motion, even though it can be noted that to date examples of catalytic rotaxanes far outnumber catalytic catenanes. Looking forward, it is possible to envisage further fundamental developments and one hopes real-world application of discoveries in this field.