Romain Jamagne
a,
Martin J. Power
a,
Zhi-Hui Zhang
b,
Germán Zango
a,
Benjamin Gibber
a and
David A. Leigh
*ab
aDepartment of Chemistry, University of Manchester, Oxford Road, Manchester M13 9PL, UK. E-mail: david.leigh@manchester.ac.uk
bSchool of Chemistry and Molecular Engineering, East China Normal University, Shanghai 200062, P. R. China
First published on 5th September 2024
The active template synthesis of mechanically interlocked molecular architectures exploits the dual ability of various structural elements (metals or, in the case of metal-free active template synthesis, particular arrangements of functional groups) to serve as both a template for the organisation of building blocks and as a catalyst to facilitate the formation of covalent bonds between them. This enables the entwined or threaded intermediate structure to be covalently captured under kinetic control. Unlike classical passive template synthesis, the intercomponent interactions transiently used to promote the assembly typically do not ‘live on’ in the interlocked product, meaning that active template synthesis can be traceless and used for constructing mechanically interlocked molecules that do not feature strong binding interactions between the components. Since its introduction in 2006, active template synthesis has been used to prepare a variety of rotaxanes, catenanes and knots. Amongst the metal–ion-mediated versions of the strategy, the copper(I)-catalysed alkyne–azide cycloaddition (CuAAC) remains the most extensively used transformation, although a broad range of other catalytic reactions and transition metals also provide effective manifolds. In metal-free active template synthesis, the recent discovery of the acceleration of the reaction of primary amines with electrophiles through the cavity of crown ethers has proved effective for forming an array of rotaxanes without recognition elements, including compact rotaxane superbases, dissipatively assembled rotaxanes and molecular pumps. This Review details the active template concept, outlines its advantages and limitations for the synthesis of interlocked molecules, and charts the diverse set of reactions that have been used with this strategy to date. The application of active template synthesis in various domains is discussed, including molecular machinery, mechanical chirality, catalysis, molecular recognition and various aspects of materials science.
Fig. 1 Classic examples of the passive metal template synthesis of interlocked molecules. (a) The Sauvage group's synthesis of catenane 4-Cu(I) in two sequential steps: coordination of the bidentate ligands to the metal in a tetrahedral geometry, followed by capturing of the threaded structure through covalent bond formation.7 (b) The Leigh group's synthesis of benzylic imine catenanes by imine formation about an octahedral Zn(II) template, followed by reduction of the imines and extraction of the metal ions to give wholly organic catenane 5.9 Although the Zn(II) ions promote formation of the imine bonds in addition to holding the ligands in place, the metal cannot be removed from the imine catenate without a subsequent reduction step to stabilise the covalent framework in the absence of the metal. The two-step synthesis of 5 is therefore an example of passive metal template synthesis. |
Since this seminal work, numerous strategies for positioning molecular components for interlocking have been developed, utilising diverse metal coordination modes and geometries10–12 and/or noncovalent interactions based on aromatic stacking interactions,13,14 hydrogen bonding,15 and/or hydrophobic effects16,17 (Fig. 2).18 The classic metal template approach7,9–12 to interlocked molecules involves two distinct processes which are generally accomplished step-wise (Fig. 2a):
Fig. 2 Schematic representation of (a) various common passive template strategies for rotaxane synthesis18–20 and (b) active template synthesis of rotaxanes.19,20,22 (a) The ‘clipping’ and ‘threading-and-stoppering’ approaches involve thermodynamically driven assembly processes followed by capturing of the interlocked structure by covalent bond formation.18 The ‘slipping’ or ‘slippage’ approach18 involves the macrocycle passing over the stoppers at an elevated temperature; at lower temperatures the macrocycle is then kinetically trapped on the axle. As a covalent bond does not need to be broken to disassemble such a threaded structure, such supramolecular assemblies are better described as kinetically stable pseudo-rotaxanes rather than as rotaxanes (see footnote 194 in ref. 21). (b) Active template synthesis is kinetically driven; component assembly and the covalent capture of the interlocked structure occur contemporaneously. |
(i) Component positioning: in which the well-defined coordination geometry of the metal is used to arrange ligands such that joining them together favours the formation of a mechanically interlocked structure.
(ii) Component connecting: a subsequent step in which the threaded arrangement present in the supramolecular complex is captured through one or more covalent bond-forming reactions.
As the component positioning step only exploits the coordination geometry of the metal ion, rather than any ability of the metal to promote the formation of covalent bonds that persist once the metal has been removed, this approach can be categorised as ‘passive’ metal template synthesis.10 In contrast, ‘active’ metal template synthesis19,20,22 (Fig. 2b) proceeds in a single step that exploits both the coordination geometry of a metal (in acting as a template to organise ligands into a threaded or entwined intermediate), and the metal's ability to catalyse covalent bond-forming reactions (in order to covalently capture the interlocked structure).
Passive and active template methodologies diverge in the principles that underpin them (Fig. 2).19 The passive template approach generally involves the formation of a thermodynamically favourable threaded or entwined intermediate complex (Fig. 2a).18 This minimum energy arrangement of building blocks is subsequently captured, typically by macrocyclisation or stoppering, in a second, bond-forming, process. In contrast, active template synthesis (Fig. 2b) does not involve a stable intermediate, but rather depends upon a catalysed bond-forming reaction occurring faster through a macrocyclic cavity than outside of it, irrespective of whether any attractive noncovalent interactions between the components are present in the final product.19,20 The active template process is thus kinetically driven and offers the prospect of assembling thermodynamically unfavourable interlocked structures.
Active template synthesis offers a number of distinctive features and potential advantages compared to passive template strategies:
(i) The absence of recognition motifs between the components of the final product provides an opportunity for the traceless synthesis of so-called ‘impossible’ rotaxanes.23–26
(ii) Sub-stoichiometric amounts of the metal can often be used,22,27 which turn over during their catalysis of the covalent bond-forming reaction(s).
(iii) The formation of a sterically shielded threaded structure can kinetically stabilise otherwise transient intermediates, aiding the elucidation of reaction mechanisms.28
(iv) The kinetically driven nature of active template synthesis allows for the formation of kinetically stable (because of the mechanical bond), but thermodynamically unstable, interlocked structures.19,20
Fig. 3 Introduction of active metal template synthesis by Leigh and co-workers.22 |
In 2011, the Goldup group found that smaller bipyridine macrocycles gave higher [2]rotaxanes yields, particularly with short axles.29 The Crowley group subsequently demonstrated that for exo-alcohol-functionalised pyridyl macrocycles, even smaller macrocycles could be used,30 including 24-membered rings which begin to approach the limit possible31 for accommodating the thickness of a threaded saturated hydrocarbon chain.
In the early-2000s various intermediates for the CuAAC reaction had been postulated in experimental and computational studies.33–35 However, unambiguous evidence for various intermediates, and in particular whether one or two Cu(I) atoms were involved, remained unclear. In 2007 Leigh and co-workers reported unexpected [3]rotaxane formation during active metal template synthesis with a pyridine macrocycle (Fig. 4).28 The trapping of two rings threaded onto an axle produced by the formation of one triazole group, using macrocycles that could clearly bind to only one metal ion each, provided clear experimental evidence that two distinct Cu(I) atoms are involved in catalysing the [3+2] cycloaddition, at least with this type of ligand system.
Fig. 4 Mechanism of the active template CuAAC reaction leading to [2]- or [3]rotaxanes depending on the ratio of macrocycle to Cu(I).28 |
In the proposed reaction pathway (Fig. 4), macrocycle 6 and the half-axle components bind to Cu(I) to form 7-Cu(I). This intermediate then undergoes the Cu(I)-catalysed 1,3-cycloaddition leading to interlocked complex 8-Cu(I) and, ultimately, metal-free [2]rotaxane 9. Alternatively, the second metal ion of 10-[Cu(I)]2 can be bound to another macrocycle prior to the azide–alkyne cycloaddition occurring, leading to [3]rotaxane 11. The formation of [3]rotaxane is enhanced by increasing the macrocycle:Cu(I) ratio, consistent with this mechanistic interpretation. Goldup and coworkers later found that high temperatures, non-coordinating solvents and larger macrocycles can also increase the amount of [3]rotaxane formed.36
Goldup and coworkers have isolated a Cu(I)-triazolizide rotaxane complex from an active template CuAAC reaction.37 The complex was remarkably stable, even under aqueous conditions, with the tightly threaded macrocycle sterically inhibiting protonation of the Cu–C bond. This was the first triazolide intermediate isolated from a CuAAC reaction.
These findings illustrate well how mechanical bonding can provide mechanistic insights into the catalysis manifolds used for active template synthesis.
Leigh and co-workers have assembled a trefoil knot 12 using a combination of passive and active template processes (Fig. 5).45 The trefoil knot is formed in 24% yield from a single molecular strand, using one Cu(I) ion to coordinate to both bipyridine units, forming a loop in the strand, while a second Cu(I) ion acts as the active template site, catalysing the CuAAC reaction of the alkyne and azide termini through the loop.45
Fig. 5 Molecular trefoil knot 12 synthesised through active template CuAAC synthesis through the macrocyclic loop formed by Cu(I)-coordination to the two bipyridine groups.45 |
Fig. 6 (a) Proposed mechanism for the active template synthesis of [2]rotaxanes through the Cu(I)-mediated Glaser–Hay homocoupling of terminal alkynes.46 (b) Selected examples of interlocked structures synthesised by active template Glaser couplings.47–49 (c) Saito's post-assembly modification a diyne rotaxane to an aryl pyrrole rotaxane.50 |
Active template Glaser and Glaser-type couplings have subsequently been used to synthesise a variety of interlocked molecules, including [2]rotaxanes,47,50–64 a [2]catenane,65 and doubly threaded [3]rotaxanes such as 18 (Fig. 6b).48,66 The approach can be combined with passive template approaches to create more elaborate architectures such as 19.49 Its suitability for forming π-conjugated rotaxanes has been utilised to synthesise an array of ‘molecular wires’ (see section on sheathed ‘molecular wires’).
Although most active template Glaser couplings use Cu(I)-phenanthroline complexes, an exception is [2]rotaxane 17 (Fig. 6b), synthesised from a macrocyclic Hamilton-type receptor, albeit in modest (9%) yield.67
Saito and co-workers have investigated the effects of macrocycle size, axle length and stoichiometry on the yields of singly threaded [2]- and doubly threaded [3]rotaxanes made via active template Glaser couplings. The ratio of [3]rotaxane to [2]rotaxane was increased using larger macrocycles and longer thread building blocks.48
Interlocked molecules prepared by Glaser couplings are readily modified by reaction of the diyne, increasing the diversity of structures accessible using this active template reaction. Saito and co-workers described the post-modification of rotaxane 20 to [2]rotaxane 21 bearing an aryl pyrrole moiety in the axle (Fig. 6c).50 This allowed the shuttling dynamics of rotaxanes with different levels of steric bulk in the axle to be studied,54 as well as chiral HPLC separation of mechanically planar chiral rotaxanes with rotationally asymmetric macrocycles.53
The Cadiot–Chodkiewicz active template synthesis of [2]rotaxanes was introduced by Leigh and co-workers in 2008 (Fig. 7).67 This reaction can be particularly sensitive to the order in which the building blocks, reagents and catalyst are combined: mixing a terminal alkyne, alkyne halide and bipyridine macrocycle 22 with an amine base and CuI resulted in a mixture of desired heterocoupling and undesired homocoupling products.67 In contrast, preforming copper acetylide 23-Cu(I), followed by subsequent addition of the macrocycle and alkyne halide 24 increased the rotaxane yield up to 85% with almost exclusive selectivity for the heterocoupled [2]rotaxane 25. The mechanism appears to proceed through the formation of Cu(I)-acetylide complex 26-Cu(I), which undergoes oxidative addition of 24 through the opposite face of the macrocycle to form 27-Cu(III). Reductive elimination affords the heterocoupled product, rotaxane 25. The strategy was later extended to the active template synthesis of heterocircuit [2]catenanes.42
Fig. 7 Proposed mechanism for the active template synthesis of [2]rotaxanes through the Cu(I)-mediated Cadiot–Chodkiewicz heterocoupling of terminal alkynes and alkyne halides.67 |
The Cadiot–Chodkiewicz active template reaction has been used for the synthesis of a variety of interlocked structures,68–71 often with better yields than the comparative Glaser couplings or enabling the formation of rotaxanes that were not accessible through alkyne homocouplings.68 It has proved particularly well-suited for the synthesis of polyyne and cumulene [2]rotaxanes, which generally show improved stability over their non-interlocked equivalents (see section on sheathed ‘molecular wires’).
Jasti and coworkers have made an impressive range of fluorescent ‘nanohoop’ [2]rotaxanes and [c2]daisy-chain rotaxanes through Cadiot–Chodkiewicz active template synthesis. The structures feature a shape-persistent macrocyclic belt containing a 2,6-connected pyridine residue (e.g. 26) to endotopically coordinate the metal that promotes rotaxane formation. The [c2]daisy-chain rotaxane 27 has a relatively compact structure, apparent from the space-filling representation of the X-ray structure (Fig. 8).69,70
Fig. 8 (a) Jasti's synthesis of a [c2]daisy-chain rotaxane 27 from an active template Cadiot–Chodkiewicz reaction. (b) Space-filling representation of 27.69 |
Fig. 9 Saito's active template synthesis of a [2]rotaxane by Ullmann C–S coupling of a thiol and aryl iodide.46 |
The same Cu(I)-phenanthroline macrocycle has also been used in a Sonogashira-type active template heterocoupling of aryl iodides and terminal alkynes, generating rotaxanes in up to 52% yield (Fig. 10).72 Concurrent oxidative homocoupling of the terminal alkynes occurs in such reactions and so the corresponding diyne rotaxanes were formed as side-products in each case. The mechanism is presumed to proceed via the formation of Cu(I)-acetylide 31-Cu(I), followed by oxidative addition of aryl iodide 32 to form Cu(III)-complex 33-Cu(III). The latter undergoes reductive elimination through the macrocycle cavity to form [2]rotaxane 34. The best yields and selectivity for the heterocoupling product were obtained using the ortho-isomer of 32.
Fig. 10 Saito's active template synthesis of [2]rotaxanes through the Cu(I)-mediated Sonogashira-type coupling of terminal alkynes and aryl iodides.72 |
Leigh and co-workers have used Cu(I) catalysis in the active template Goldberg amidation of aryl halides to form [2]rotaxanes with chiral C2-symmetric cyclohexyldiamine macrocycles.73 The mechanism is similar to the active template Cu(I)-mediated Sonogashira-type coupling (Fig. 10), with an aryl amide rather than an alkyne undergoing ligation with 21-Cu(I). The resulting chiral rotaxanes were subsequently investigated as ligands for enantioselective metal catalysis (see section on asymmetric, diastereoselective and other types of catalysts).
Leaning on previously reported Pd(II)-mediated passive template syntheses of rotaxanes78–81 and catenanes for the macrocyclic ligand design,82,83 Leigh and co-workers showed that, despite its square planar coordination geometry, Pd(II) can be a suitable metal for the active template synthesis of [2]rotaxanes through alkyne homocoupling (Fig. 11).84 The reaction proceeds with the replacement of the trans-chloride ligands of 35-Pd(II) by the Cu(I)-acetylide unit 36 to form a threaded trans-37-Pd(II) complex. Isomerisation to cis-38-Pd(II) occurs without breaking the Pd–acetylide bond, preserving the threaded architecture. Reductive elimination occurs from the cis configuration to yield diyne [2]rotaxane 39 after decomplexation of the weakly coordinating Pd(0). Only a catalytic amount of Pd(II) (5 mol%) is required, together with I2 or O2 to reoxidise the Pd(0) formed in situ, giving rotaxane yields up to 90%.
Fig. 11 Active template synthesis of [2]rotaxanes through the Pd(II)-mediated oxidative homocoupling of terminal alkynes.84 |
The oxidative Heck cross-coupling reaction85 is a variant of palladium cross-couplings that does not feature Pd(0) in the mechanism. Accordingly, it is more suitable as the basis for active template synthesis. Leigh and co-workers demonstrated its efficacy in the synthesis of a variety of [2]rotaxanes using a bipyridine macrocycle and boronic acid and electron-poor alkene axle building blocks (Fig. 12).27 The catalytic loading could be as low as 1 mol% while still furnishing a rotaxane yield of 66%, the largest catalytic turnover number reported to date for an active template reaction.
Fig. 12 Active template synthesis of [2]rotaxanes through Pd(II)-mediated oxidative Heck cross-coupling.27 |
The examples in Fig. 11 and 12 demonstrate the significant potential of palladium for active template synthesis, particularly if macrocyclic ligands can be found that can overcome the lability of Pd(0) so that it remains coordinated to the components during various important steps in the mechanism.
Very high yielding four-component synthesis of rotaxanes by successive Pd(II)-promoted Michael additions of α-cyano esters to alkyl vinyl ketones has been reported using tridentate pyrroloimidazolone or bisoxazoline pincer macrocycles (Fig. 13).86 The formation of complex 40-Pd(II) enables two consecutive 1,4-conjugate additions to α,β-unsaturated ketone 41 from opposite macrocycle faces to quantitatively form [2]rotaxane 42-Pd(II). In this example the template interactions between thread and macrocycle remain in the rotaxane product, a rare feature for active template synthesis. Accordingly, the intercomponent interactions were used to good effect for the synthesis of rotaxanes with well-defined switching of the macrocycle position on the axle (molecular shuttles).
Fig. 13 Active template synthesis of [2]rotaxanes through successive Pd(II)-mediated Michael additions of four components.86 |
Active template synthesis that deletes atoms from the building blocks while joining them to form a hydrocarbon region of a threaded axle is a particularly appealing target because it is completely traceless. The main obstacle to doing so lies in achieving effective and catalytic formation of C(sp3)–C(sp3) bonds using organometallic chemistry. This is generally difficult due to slow rates of oxidative addition of metal centres to C(sp3)–heteroatom bonds, concomitant with facile β-hydride elimination of the resulting organometallic intermediate.88–91 Most successful C(sp3)–C(sp3) coupling protocols involve alkyl halides activated towards oxidative addition and lacking β-hydrogens.92,93
Adapting the work of Fu and co-workers on the coupling of unactivated alkyl halides,94–98 the active template synthesis of alkyl chain axle-region rotaxanes was achieved by Leigh and co-workers through the Ni(0)-catalysed C(sp3)–C(sp3) homocoupling of bromoalkanes (Fig. 14).99 The reaction was initially thought to likely proceed through a Negishi-type mechanism,91 however mechanistic studies indicate a pathway similar to the Ni-mediated coupling of aryl halides.100 Oxidative addition of alkyl bromide 43 to 44-Ni(0) is followed by the one-electron reduction of Ni(II), allowing a second oxidative addition of 43, ligating the axle building blocks through opposite faces of the macrocycle. Reductive elimination then captures threaded architecture 44, while Zn(0) reduces the resulting Ni(I) species to catalytically active Ni(0).
Fig. 14 Active template synthesis of traceless [2]rotaxanes through the Ni(0)-mediated homocoupling of unactivated alkyl bromides.99 |
Saito and co-workers have carried out the cross-coupling of iodoarenes mediated by a dibenzodihydrophenanthroline–Ni complex.101 This was used for the synthesis of [2]rotaxanes with up to ten para-phenylene units in the axle.
Active template synthesis by Ni-catalysed homocouplings of alkyl bromides can produce multiply threaded rotaxanes. Leigh and co-workers isolated doubly threaded rotaxane 45 in up to 51% yield using a terpyridine 35-membered macrocycle (Fig. 15a).102 During the process, [2]rotaxane 44 was also produced but the ratio of doubly- over singly-interlocked product could be increased by using 43 in large excess. The mechanism appears to proceed in a stepwise manner in which the [2]rotaxane is formed first, followed by binding of a second set of axle building blocks to the metal coordinated in the [2]rotaxane to yield [3]rotaxane 45 after a second reductive elimination.
Fig. 15 (a) Mechanism of the active template Ni(0)-mediated active template synthesis of rotaxanes with one, two or three axles threaded through a single macrocycle. (b) Doubly and triply threaded rotaxanes formed using this method.102 |
A triply threaded [4]rotaxane 46 was obtained using the same strategy by increasing the size of the macrocycle from a 35-membered ring to either a 37- or a 38-membered ring (Fig. 15b).103 The increased size meant the macrocycle could accommodate the formation of a third axle within the cavity. [2]Rotaxane was not observed in the product distribution using the larger ring systems, only the doubly- and triply-threaded products. This suggests that the [2]rotaxane does not have stoppers large enough to prevent dethreading of the larger macrocycles unless the metal remains coordinated (which reduces the effective size of the cavity, and the coordinated metal can then promote the formation of the second and third axle; Fig. 15a). Also consistent with this, when a slightly larger 39-membered macrocycle was used only non-interlocked coupled products were observed. To date, rotaxane 46 is the sole example of a rotaxane containing more than two axles threaded through a single macrocycle.
The formation of traceless rotaxanes by active template synthesis was later extended by the Leigh group to the preparation of unsymmetrical axle [2]rotaxanes by the Ni-catalysed C(sp3)–C(sp3) cross-coupling of alkylzinc with redox-active esters (Fig. 16),104 a reaction introduced by Baran and co-workers.105 Transmetallation of Ni(I)-bipyridine macrocycle 47-Ni(I) with alkylzinc 48, followed by single electron transfer (SET) from the Ni(I) to the phthalimide group of redox-active ester 49 and consecutive decarboxylative fragmentation, leads to cationic intermediate 50-Ni(II).106 Radical recombination renders Ni(III) intermediate 51-Ni(III), which undergoes reductive elimination through the macrocycle cavity to form [2]rotaxane 52, while regenerating the catalytic Ni(I) species. [2]Rotaxane 53 was also formed, resulting from the active template homocoupling of alkylzinc 48.
Fig. 16 Mechanism of the active template synthesis of traceless [2]rotaxanes through the Ni(I)-mediated heterocoupling of alkylzinc and redox-active esters.104 |
Recently, Kimizuka, Yagi and Itami reported a novel Ni-mediated homocoupling active template reaction employing a similar bipyridine macrocycle in the synthesis of catenane 54 (Fig. 17).107 The threading unit is based on Itami's previous synthesis of catenanes and a trefoil knot through a covalent scaffold approach,108 and features cyclohexa-1,4-diene units109 that are converted to phenyl groups to form the all-benzene ‘nanobelt’ of the final catenane.
Fig. 17 Kimizuka, Yagi, and Itami's Ni(0)-mediated active template synthesis of [8]cycloparaphenylene catenane 54.107 |
Fig. 18 Active template synthesis of [2]rotaxanes through the Zn(II)-mediated Diels–Alder of imidazolidones and cyclopentadiene derivatives.110 |
A small number of active metal template reactions have been developed that proceed through radical mechanisms involving a redox process between the metal and the ligand. Co(II)porphyrinates form nucleophilic radical carbenes that react with olefins to form cyclopropanes.111–113 Megiatto and co-workers exploited this chemistry for the active template synthesis of rotaxanes with macrocycle 60-Co(II).114 The mechanism involves a radical carbene transfer reaction between diazo and styrene half-threads (Fig. 19). Coordination of 3,5-diphenylpyridine ligand 61 to the exo-face of the Co(II)porphyrinate (it is too bulky to coordinate to the endo-face) causes diazo-acetate 62 to react with macrocycle 63-Co(II) inside the cavity, leading to Co(III)-carbene radical 64-Co(III). Subsequent carbene transfer with styrene derivative 65 generates rotaxane 66-Co(II). In the absence of 61 the yield of [2]rotaxane decreased from 95% to 74%,115 demonstrating the importance of coordination of an exo-ligand for promoting formation of the interlocked product over the free thread.
Fig. 19 Megiatto's active template synthesis of [2]rotaxanes through the Co(II)-mediated radical carbene transfer between diazo and styrene derivatives.114 |
The same group reported that a Ru(II)-porphyrinate macrocycle promotes N–H carbene insertion between amines and diazo half-threads in an active template fashion (Fig. 20).116 Ru(II)-porphyrinate 67-Ru(II) differs from 60-Co(II) in that it is only composed of aromatic C–H bonds and therefore chemically inert to Ru(II)-carbenoids. Introduction of a bulky carbene ligand exo- to the macrocycle forces reaction of diazo 69 to occur inside the cavity. Addition of amine 68 generates [2]rotaxane 70-Ru(II) in quantitative yield.
Fig. 20 Megiatto's active template synthesis of [2]rotaxanes via Ru(II)-mediated N–H carbene insertion.116 |
The distinctive reactivity and catalytic properties of rhodium have been exploited for active template synthesis by the Chaplin group.117–120 They first described the formation of pseudorotaxane 71-Rh(I) by a Rh(I)-promoted terminal alkyne coupling reaction that forms an enyne axle (Fig. 21).117 Ligand exchange between terminal alkyne 72 and Rh(I) ethylene complex 73-Rh(I) leads to Rh(I)-alkyne complex 74-Rh(I), which undergoes oxidative addition to give 75-Rh(III). A second equivalent of 72 inserts into the Rh–H bond in a ‘head-to-tail’ type mechanism to form a Rh(I) gem-alkenyl complex 76-Rh(III) that is sufficiently stable to be isolated. However, upon heating to 80 °C, 76-Rh(III) reductive eliminates and does not afford the expected Rh(I) gem-enyne pseudorotaxane 77-Rh(I), but instead quantitatively forms E-enyne pseudorotaxane 71-Rh(I). This suggests a mechanism in which β-hydride abstraction of 76-Rh(III) occurs and the equilibrium is driven towards the formation of E-enyne Rh(III) complex 78-Rh(III), which can rapidly undergo reductive elimination to form 71-Rh(I). Only the gem-enyne product is formed when the reaction is performed with an acyclic CNC ligand. This study illustrates how active template synthesis can provide insights on ligand tuning to change reaction regioselectivity.118
Fig. 21 Chaplin's active template synthesis through the Rh(I)-mediated homocoupling of terminal alkynes to form a threaded E-enyne.117,118 |
The Chaplin group developed a Rh-mediated active template synthesis of diyne rotaxanes from alkynyl Grignard reagents (Fig. 22).119 Transmetallation of alkynyl Grignard 79 with a macrocyclic phosphinite pincer complex 80-Rh(III) leads to 81-Rh(III). Subsequent decarbonylation promoted by Me3NO affords diyne [2]rotaxane 82-Rh(I) through reductive elimination (81% yield over 2 steps). The group later reported a similar pincer complex with bulkier stoppers that could be converted to a hydrocarbon-based rotaxane.120 Demetallation followed by conversion of the phosphine ligands to thiophosphenes gave [2]rotaxanes, while a subsequent clipping reaction was used to synthesise a [2]catenane.
Fig. 22 Chaplin's active template synthesis of diyne rotaxanes through Rh(III)-mediated homocoupling of alkynyl Grignard reagents.119 |
Fig. 23 Mock's synthesis of a [2]rotaxane through cucurbituril-promoted Huisgen azide–alkyne cycloaddition of ammonium-functionalised axle building blocks.121 |
Similar complexation-driven effective molarity increases to accelerate axle-forming reactions have been used to assemble (poly)rotaxanes122 and for ‘cooperative capture’ synthesis.123 In addition to cucurbiturils, other 3D ‘molecular barrels’, such as cyclodextrins, pillararenes and calixarenes, have been used in such approaches. For example, Arduini, Credi and co-workers reported that a tri-urea-calix[6]arene accelerates the alkylation of a complexed pyridylpyridinium guest to form a rotaxane (Fig. 24).124,125 Macrocycle 84 forms a 1:1 complex with 85 in toluene, and upon addition of stoppering components 86 and 87, [2]rotaxane 88 is formed in up to 70% yield. The macrocycle accelerates alkylation of the pseudorotaxane 16× faster than the axle is alkylated in the absence of the macrocycle. The authors ascribe the rate enhancement to the increase of pyridine nucleophilicity inside the electron-rich cavity of 84 which contains multiple aromatic rings, as well as urea groups that could stabilize the transition state. No formation of 89 is observed due to the steric preferences of the non-symmetrical macrocycle. The authors later carried out the selective synthesis of all three possible orientational isomers of related calix[6]arene-based [3]rotaxanes.126
Fig. 24 Credi and Arduini's cooperative capture synthesis of a [2]rotaxane.124,125 |
Fig. 25 Metal-free active template synthesis of a [2]rotaxane through transition state stabilisation.127 |
The control experiments used to interrogate the mechanism of formation of rotaxane 93 included the use of analogues of macrocycle 90 that were missing either the two amide groups (intended to stabilise the forming anion in the transition state) or the oligoethylene glycol chain (intended to stabilise the forming cation in the transition state). However, although replacing the glycol units with a hydrocarbon chain led to no rotaxane formation, a small amount (5%) of rotaxane was still formed if the macrocycle amide groups were changed to esters. This suggested that crown ethers might be able to promote the alkylation of a primary amine through the crown ether cavity. This is somewhat counterintuitive as amines bind only very weakly to crown ethers and, when protonated as ammonium groups (which do bind strongly to crown ethers) then there is no lone pair with which to act as a nucleophile. Nevertheless, this is exactly what happens (Fig. 26).128
Fig. 26 Metal-free active template synthesis of rotaxanes from primary amines, electrophiles and crown ethers.128,129 |
Despite having modest binding constants for the thread building blocks, crown ethers such as 94 accelerate the reaction of primary alkyl- and benzylamines such as 95 with a series of electrophiles to form rotaxanes (Fig. 26).128 Kinetic studies show rotaxane formation with a benzylamine is ∼26× faster than the background thread-forming reaction. The process allows the reagent-less formation of crown ether rotaxanes in a matter of hours at room temperature, accomplished by simply mixing together the macrocycle and the two thread-forming building blocks in toluene.
The rotaxane formation is driven by crown ether hydrogen bonding stabilising the partial positive charges formed in the transition state, a process that has previously been observed in ethylene-glycol-oligomer-catalysed aminolysis reactions.130,131 The utility of the metal-free active template reaction was demonstrated by the one-step synthesis of (i) pH-switchable molecular shuttles via aza-Michael addition and (ii) crown ether-peptide rotaxanes via N-acylation (Fig. 26).128
The system was subsequently extended to other electrophiles, producing crown ether rotaxanes based on (thio)urea, carbamate, sulfonamide and phosphoramidate/phosphinamide-functionalised axles in up to 95% yield and >100:1 rotaxane-to-free-thread selectivity.129,132 It also proved possible to use crown ether catalysed N-arylation of primary amines with electron-deficient aryl halides to produce rotaxanes with aniline threads. This metal-free active template N-acylation strategy has been used for the single-step enantioselective synthesis of mechanically planar chiral [2]rotaxanes (see section on mechanical and topological chirality).133
In an important recent development, the principle of enhancing the nucleophilicity of primary amines by hydrogen bonding to a macrocycle has been expanded from crown ethers to cyclic peptides.134 The Leigh group have previously threaded cyclo(proline) macrocycles such as 96 onto diammonium axles to form rotaxanes of cyclic peptides by passive template synthesis in up to 63% yield.135 Numata and coworkers found that this macrocycle-ammonium passive template motif could be adapted to metal-free macrocycle-amine active template synthesis in a similar manner to the crown ether-ammonium/amine system.134 With a glycine-derived amine nucleophile and phenylalanine-derived nitrophenol ester electrophile metal-free active template synthesis generated wholly peptidic rotaxanes (Fig. 27).134 The yields reported are very modest (∼1%), which the authors ascribe to the tendency of cyclic peptide oxygens to point out of the cavity and the presence of competing intermolecular hydrogen bonds, both of which limit the ability of the macrocycle to form strong hydrogen bonds with the amine. As well as being a strategy to mechanically interlocked peptides, the extension of metal-free active template synthesis from crown ethers to ammonium-binding cyclic peptides suggests that other ammonium-binding macrocycles (e.g. cucurbiturils, pillararenes, etc) may prove efficacious at accelerating the reaction of amines through a macrocyclic cavity for active template synthesis.
Fig. 27 Numata's metal-free active template synthesis of wholly peptidic rotaxanes from a glycine nucleophile, a phenylalanine-nitrophenol ester electrophile and cyclo(proline) macrocycle.134 |
Fig. 28 Zhang's metal-free active template synthesis of protein heterocatenanes.136–138 (a) SpyStapler-mediated isopeptide bond formation between SpyTag and BDTag.137 (b) Active template synthesis of a protein [2]catenane from a cyclic protein incorporating the SpyStapler sequence (c-SpyStapler-POI) and a linear protein terminated with the BDTag and SpyTag sequences at either end.136 (c) Higher order protein [n]catenanes (n = 2–5) assembled using a mutated SpyStapler sequence (SpyStapler003).138 |
The group have also developed a novel protein (‘AT-Snoop’) for active template protein synthesis.139 AT-Snoop and AT-Spy both catalyse isopeptide bond-formation but react orthogonally using different proteins domains. The orthogonality was exploited for the synthesis of [n]heterocatenanes, including an asymmetric protein olympiadane.138
Fig. 29 Yagai's self-assembling polycatenanes consisting of ∼13 nm diameter supramolecular rings.140 The molecules with a polar head group, rigid segment and non-polar tail assemble into rosettes that stack to form helical strands and toroids. The internal surface of the toroids seed the formation of new stacks, leading to supramolecular polycatenanes.140 |
Fig. 30 Active template Cadiot–Chodkiewicz synthesis of molecular shuttle 97. The position of the macrocycle on the axle is governed by relatively weak intercomponent interactions.67 |
Molecular shuttle 97 was synthesised through an active template Cadiot–Chodkiewicz heterocoupling, with one alkyne unit containing an aniline and the other a dimethylaminopyridine (DMAP) group (Fig. 30).67 In its unprotonated form the macrocycle of 97 preferentially coordinates to the aniline site on the axle through a single hydrogen bond with an intercomponent binding energy of −3.9 kcal mol−1, which would be far too weak to thermodynamically drive rotaxane synthesis. Protonation or Li+ coordination of the DMAP group of 97 triggers the translocation of the macrocycle to the DMAP axle site. This illustrates the efficacy of active template synthesis for the preparation of rotaxane-based molecular shuttles with much weaker intercomponent interactions (and correspondingly faster dynamics) than those required for passive template methods.143
The active template Cadiot–Chodkiewicz reaction has been used to develop other mechanically interlocked structures without strong intercomponent binding interactions, such as the [c2]daisy chain rotaxane 27 described by Jasti and co-workers (Fig. 8). This structure can be switched from a comparatively stable contracted form at temperatures lower than −40 °C to an extended state at higher temperatures.59
Goldup and co-workers used active template CuAAC synthesis to prepare molecular shuttles in which the shuttling of a bipyridine ring between bidentate and tridentate coordinating sites144 could be controlled by metal ions with different preferential coordination geometries.145
Weiss, Guldi, Hayashi and co-workers have reported light-switchable molecular shuttling in a [2]rotaxane based on an electron donor–acceptor system composed of a Zn–porphyrin-based macrocycle and a fullerene-stoppered thread.146 An active template CuAAC reaction installs a triazole moiety in the axle that forms one binding site in the resulting shuttle.
Beer and co-workers have also reported porphyrin-based molecular shuttles synthesised by active template protocols. In those examples the position of a pyridine macrocycle (used to direct the CuAAC reaction) is controlled via the addition of a Lewis base or anion.147
Fig. 31 Goldup's active template synthesis of mechanically planar chiral rotaxanes (Smp)- and (Rmp)-102.148 |
The approach was later expanded to diastereoselective synthesis of mechanically planar chiral rotaxanes by modifying the stereodirecting moiety in the half-thread building block (Fig. 32).149,150 Excellent diastereomeric ratios of 98:2 were obtained using chiral amino acid-derived azide 102. Post-assembly removal of the point chirality source by symmetrisation of the amino acid stereocentre rendered the enantioenriched mechanically planar chiral rotaxane (Smp)-103. Similar methodology was later applied to the stereoselective preparation of enantiopure topologically chiral catenanes.151
Fig. 32 Goldup's diastereoselective active template synthesis of mechanically planar chiral rotaxane (Smp)-103.149 |
The Goldup group have also demonstrated that this chiral auxiliary can act as an mechanical interlocking auxiliary,23–25 i.e. a group that facilitates the formation of the mechanical bond but is subsequently cleaved from the structure after shuttling of the macrocycle to a different part of the axle (Fig. 33).25 The auxiliary was removed to leave ‘impossible’ rotaxane23,24 (Smp)-104, with no recognition motif between the axle and macrocycle. The group also investigated attaching the chiral auxiliary to the macrocycle, rather than an axle-forming component.152 This could be used to direct the formation of both mechanically planar chiral rotaxanes and topologically chiral catenanes with good stereoselectivity (typically >90% ee).
Fig. 33 Goldup's mechanical interlocking chiral auxiliary for the active template synthesis of ‘impossible’ mechanically planar chiral rotaxanes.25 |
CuAAC active template synthesis has also been used by the Papot group for the diastereoselective synthesis of [1]rotaxanes (‘molecular lassos’).153,154 The point-chiral stereogenic unit has not yet been removed, meaning that a mechanically planar chiral [1]rotaxane with chirality solely arising from the mechanical bond remains to be achieved.
The active template mechanical interlocking chiral auxiliary systems developed by the Goldup group have been used to make a number of catenanes and rotaxanes with other, previously unexplored, types of chirality, including co-conformationally ‘pseudo-topologically’ chiral catenanes,155 mechanically (but not topologically) planar catenanes,156 mechanically axially chiral catenanes and noncanonical mechanically axially chiral rotaxanes.157–159 The topological aspects of these fascinating systems has recently been reviewed elsewhere.160 The versatility and efficiency of the active template approach has facilitated the exploration of new types of chiral interlocked molecules, leading to a better understanding of aspects of molecular asymmetry.
The Leigh group reported a single-step asymmetric synthesis of mechanically planar chiral rotaxanes by using primary amine 105, activated ester 106a–b with a chiral leaving group and achiral crown ether macrocycle 107 with Cs symmetry in an application of metal-free active template synthesis (Fig. 34).161 The incorporation of cinchonidine pseudoenantiomers as directing groups enabled the synthesis of either enantiomer of [2]rotaxane 108 in 40–50% ee. X-Ray crystal structures and computational modelling indicated π-stacking between the naphthalene ring of the macrocycle and the electron-deficient aromatic stopper helps to direct the reaction stereoselectivity.
Fig. 34 Single-step asymmetric metal-free active template synthesis of mechanically planar [2]rotaxanes.161 |
A different approach to mechanically planar chiral rotaxanes, also exploiting metal-free active template synthesis, was developed by Tian and Zhu.162 Modified bis(chloroaryl) crown ethers were used as catalysts for the spontaneous formation of [2]rotaxanes using a range of electrophiles and amine nucleophiles. The prochiral interlocked molecule (with an unsymmetrical axle but a rotationally symmetric ring) then underwent catalytic desymmetrisation through a Pd-catalysed asymmetric Suzuki–Miyaura reaction to yield the chiral rotaxanes in up to 97:3 er.
Fig. 35 Mechanically chiral [2]rotaxane 109, synthesised through a Goldberg active template reaction, and its scope as a ligand for the nickel-catalysed enantioselective Michael addition of diethyl malonate and trans-β-nitrostyrenes.73 |
The Goldup group have demonstrated that chiral rotaxanes can be used as ligands for diastereoselective catalysis. An active template CuAAC reaction was used to synthesise the interlocked ligand for [2]rotaxane-gold catalyst 113-Au(I).163 The catalytic activity could be switched on by the addition of catalytically innocent metal cations (Fig. 36). 113-Au(I) was shown to catalyse Toste's Au(I)-mediated modification of the Ohe–Uemura cyclopropanation reaction.164,165
Fig. 36 Goldup's [2]rotaxane precatalyst 113-Au(I) prepared by an active template CuAAC reaction, and the effect of additives on its catalytic efficacy in the Toste–Ohe–Uemura cyclopropanation reaction.163 |
Building on this system, Goldup and co-workers described a catalyst for asymmetric synthesis that relies on the mechanically planar chirality of the rotaxane ligand as the sole source of chiral induction.166 Ligand (Rmp)-114 was synthesised via a diastereoselective active template CuAAC reaction149 (Fig. 37a) (see section on mechanical and topological chirality). Upon coordination to Au(I) and activation with a cofactor, (Rmp)-114-Au(I) proved an effective catalyst for the Au(I)-mediated Toste–Ohe–Uemura cyclopropanation, generating a range of cyclopropanes (e.g. 115) in good yields and stereoselectivities (62–92% de for the cis diastereomer, 9–77% ee; Fig. 37b). Modelling and transition state calculations suggest that (Rmp)-114-Au(I) possesses a well-defined and rigid chiral environment around the metal centre (Fig. 37c).
Fig. 37 (a) Goldup's active template synthesis of mechanically planar chiral rotaxane precatalyst (Rmp)-114-Au(I). (b) Example of a diastereoselective and enantioselective Toste–Ohe–Uemura cyclopropanation reaction promoted by 114-Au(I). (c) Computationally modelled transition state.166 |
The Goldup group have also reported [2]- and [3]rotaxane organocatalysts, synthesised via CuAAC active template reactions, that promote the anion–π catalysed Michael addition of malonic acid monothioester 116 to β-nitrostyrene 117 (Fig. 38). The selectivity of 62:1 with [3]rotaxane catalyst 118 for addition/decarboxylation product 119 was comparable with the most effective non-interlocked catalysts previously reported.167 Computational studies suggested that the high selectivity observed may be due to π-stacking between the protonated bipyridine group and the naphthalene diimide.
Fig. 38 Goldup's [3]rotaxane anion–π organocatalyst 118 for the Michael addition of malonic acid monothioester to β-nitrostyrene.167 |
In catalysis applications other than stereoselective catalysis, Schmittel and co-workers have exploited the triazole units formed during CuAAC active template reactions to create binding sites for an allosterically regulated silver(I) catalyst.168 Catalysis by the [2]rotaxane was regulated by steric restrictions on the conformational mobility of the components inhibiting the formation of catalytically active species. They have also described the design of a three-input logic AND gate with catalytic output using a [2]rotaxane shuttle prepared by active template synthesis.169
Fig. 39 (a) Beer's active template synthesis of [2]rotaxane 120-Re(I) for anion sensing. (b) Other examples from the Beer group of halogen- and chalcogen-bonding rotaxane hosts prepared by active template synthesis.170,179,180,182 |
The Beer group have also demonstrated that Cu(I) binding to alkyl-selenium or alkyl-tellurium moieties in a macrocycle can be used to direct CuAAC active template synthesis through the macrocycle to form chalcogen-derivatised rotaxanes such as 123a–b (Fig. 39b).182 The resulting rotaxanes bind hard and soft anions in a broad range of environments.
A range of hydrogen bonding motifs unrelated to the active template assembly process have been incorporated into mechanically interlocked hosts for the recognition of small molecules and anions. McClenaghan, Tucker and co-workers prepared [2]rotaxane systems containing barbital recognition sites, such as 17 (Fig. 6b), via active template copper-catalysed CuAAC and Glaser reactions.47,183 Beer et al. constructed a neutral [2]rotaxane host for anionic guests using a hydrogen bond donating indolocarbazole unit as part of the axle.184 A [2]rotaxane host including an amide binding site was used by the Ghosh group for the electrochemical sensing of anions.185 Jasti and co-workers used a Cu-catalysed active template reaction for a family of [2]rotaxanes in which fluoride-70 or thiolate-186 induced dethreading results in the switching on of fluorescence.
The Goldup group have used CuAAC active template synthesis to build a number of rotaxanes for anion187 and cation188–190 binding. Some of the interlocked ligands stabilise distorted metal coordination geometries with unusual coordination numbers due to the sterically restricted 3D shape of the binding pocket.189 This is exploited in a Pt(II)-rotaxane complex, in which the metal is stabilised towards oxidation and acts as a reversible stimuli-responsive phosphorescent sensor for Ag(I) ions.190
Fig. 40 Papot and Leigh's active template CuAAC synthesis of β-galactosidase-cleavable [2]rotaxane 124, which releases paclitaxel in tumour cells.191 |
Besides providing temporary shielding of an axle, rotaxane formation can be used to transiently modify the solubility and absorption properties of molecular components. Lewis et al. reported the use of an a CuAAAC active template [2]rotaxane as a triggerable cage for G-quadruplex (G4) DNA binders.192 The [2]rotaxane has better cell permeability than the analogous non-interlocked DNA binder, but lower cytotoxicity (probably due to the macrocycle inhibiting access to the G4 DNA binder). Enzymatic or light-triggered cleavage of one of the rotaxane stoppers liberates the G4 DNA binder within the cell, restoring its binding properties and cytotoxicity with high spatiotemporal control. Goldup, Tavassoli and coworkers have designed a triggered Curtius rearrangement immolative system where macrocycles can be functionalised at a late stage to incorporate a range of different cargos and trigger units.193 The approach was demonstrated with bipyridine macrocycles for CuAAC active template reactions and, notably, with crown ethers for metal-free active template synthesis.
The Anderson group have used active template Glaser couplings for the synthesis of π-conjugated rotaxanes with Zn(II)-porphyrin stoppers at either end (Fig. 41).60 Porphyrin [2]rotaxane 128 was elaborated into [4]catenane 129 in 62% yield using 130 as a radial template for the interlocked architecture. [7]Catenane 131 was also formed as a Vernier template by-product of the synthesis.
Fig. 41 Anderson's assembly of [4]- and [7]catenanes from [2]rotaxanes prepared by active template Glaser homocouplings.60 |
The Anderson and Jasti groups have both used masked phenyl and alkyne units to extend the number of π-conjugated units in the axle and increase the diversity of accessible polyyne rotaxane and catenane structures. Jasti and co-workers used cyclic precursors to [n]cycloparaphenylene nanohoops for the synthesis of π-conjugated rotaxanes and catenanes via active template Cadiot–Chodkiewicz heterocouplings (Fig. 42).194 The bent cyclohexadiene moieties in 132 facilitate endotopic metal binding, which is difficult to achieve with cylindrical [n]cycloparaphenylenes. Reductive aromatisation of the cyclohexadienes furnishes the nanohoops in [3]rotaxane 133 and catenanes 134–135. The group later expanded the methodology to CuAAC active template synthesis, producing triazole catenanes in high yields.195
Fig. 42 Jasti's assembly of interlocked nanocarbons via active template Cadiot–Chodkiewicz synthesis.194 |
Anderson and co-workers have employed indane196 and bicyclo[4.3.1]decatriene197 as photolabile masked-alkyne equivalents as well as dicobalt carbonyl complexes that can be unmasked by oxidative decomplexation.198 These synthons have been used in active template Glaser and Cadiot–Chodkiewicz syntheses of rotaxanes and catenanes, including polyyne [3]rotaxane 136 (Fig. 43) which has 14 contiguous alkyne units in the axle stabilised by the nanohoop sheath. The strategy was later used to prepare a polyrotaxane with an extraordinary 34(!) contiguous triple bonds in the axle and four threaded macrocycles.199 This polyyne rotaxane is in a length regime where the electronic properties of polyynes converge to those of carbyne. The rotaxane displayed electronic properties similar to those predicted for an infinite carbyne, illustrating the potential of polyyne rotaxanes to act as molecular wires.
Fig. 43 Anderson's extended polyyne rotaxane 136 synthesised via a Cadiot–Chodkiewicz active template reaction followed by oxidative unmasking of the bridging alkynes.198 |
The synthesis of catenanes with this method has also been reported, however, unmasking of the alkynes resulted in a complex mixture of products that could not be purified.
The CuAAC active template reaction has been used by the groups of Ngo, D’Souza and Goldup for the synthesis of porphyrinoid rotaxanes.200,201 These compounds displayed similar photophysical properties to the free thread component, but the threaded macrocycles inhibit intermolecular interactions between axles including π-stacking-driven aggregation and ligand-driven dimerisation.
Leigh and co-workers have used metal-free active template synthesis to make compact amine-crown ether rotaxanes that exhibit exceptional basicity (Fig. 44).203 The super-basicity results from three main factors: (1) the formation of strong hydrogen bonding and electrostatic interactions in the protonated salt of the rotaxane; (2) the forced proximity (and lack of conformational and co-conformational freedom) of the amine and ethylene glycol functional groups, resulting in electrostatic repulsion between the interlocked macrocycle and thread; (3) a high degree of preorganisation of the components in the free-base rotaxane that favours amine protonation.
Fig. 44 Spontaneous assembly by metal-free active template synthesis of compact rotaxane superbase 137.203 |
A range of compact rotaxane superbases was prepared by metal-free active template in 1–3 steps from commercially available starting materials. The compact rotaxane superbases had pKaH+ values as high as 32.2 for 137 in acetonitrile (13 pKaH+ units higher than the non-interlocked components, substantially larger than that of typical crown ether-amine rotaxanes), which are comparable to some phosphazene superbases. The highest basicities were generated with the smallest macrocycle (21-crown-7) and the most electron-rich dibenzylamine axles. The most efficacious compact rotaxane superbases showed minimal nucleophilicity and excellent hydrolytic stability in comparison to conventional commercial superbases.
Rotaxane formation, either by active or passive synthesis, can also be used to tune photophysical properties. An illustration is the Goldup and Zysman-Colman groups use of CuAAC active template synthesis to prepare carbazole-benzophenone-based rotaxanes that display thermally activated delayed fluorescence (TADF).204 A large increase in photoluminescence quantum yield results from threading one or two rings around the TADF emitting unit. The interlocked structure also increases photostability of the emitters and altered their emission wavelength, shifting it slightly to the red compared to the non-interlocked thread. These properties arise from the close proximity between the macrocycle and axle and resulting weak interactions that finely tune the HOMO energy level, the singlet–triplet energy gap, and the energy gap between the charge transfer states of the emitters.
A [2]rotaxane system capable of reversible electronic energy transfer (REET) between chromophores attached to the axle and macrocycle of a rotaxane has been reported by the McClenaghan and Goldup groups.205 Delayed luminescence and long luminescence lifetimes result from the close proximity of the chromophores.
Interlocked systems accessible through active template synthesis have also been investigated as single-ion magnets. The rigid coordination environment present in compact rotaxanes was used to design cobalt(II) ligands that form complexes that show magnetic anisotropy and field-induced single-ion magnet behaviour.206
The restriction of interlocked component (co-)conformational freedom in relatively compact rotaxanes accessible by active template synthesis can manifest itself in effects observable by EPR207 and/or NMR spectroscopy.208 The restricted dynamics influences EPR resolution207 and hydrogen bonding interactions208 can be directly probed through 1H–15N multiple quantum coherence NMR experiments.
Fig. 45 Goldup's synthesis of homo[6]rotaxane 134 through iterative high-yielding active template CuAAC reactions.210 |
The Goldup and Tavassoli groups have prepared oligonucleotide-based rotaxanes by adapting click DNA ligation211,212 to active template CuAAC synthesis.213 Using a bipyridine macrocycle and oligonucleotide strands functionalised with an alkyne or azide group in the 3′ and 5′ positions in their chain termini, two different mechanically interlocked DNA strands could be prepared in good yield. The threaded macrocycle had a shielding effect on duplex formation, inhibiting DNA hybridisation and PCR amplification of the oligonucleotide rotaxanes compared to non-interlocked analogues.
Hybrid organic–inorganic interlocked structures offer an alternative class of materials. Szyszko and co-workers have used CuAAC active template synthesis for the assembly of a series of hybrid organic–inorganic rotaxanes incorporating cubic silesquioxanes as one of the stoppering elements.214 The hybrid rotaxanes demonstrated improved thermal stability in the solid state compared to wholly organic analogues.
Mullen and co-workers have reported the CuAAC active template synthesis of [2]rotaxanes on polymer resins.215 This generated higher yields of interlocked products than comparable passive template solid phase synthesis.
Fig. 46 A rotaxane-based peptide synthesiser 135, assembled through CuAAC active template synthesis.216 |
Catalysis-driven metal-free active template synthesis has been used to demonstrate the mechanism of chemically fuelled molecular pumps. Molecular pump 143 contains an activated ester electrophile that causes 24-crown-8 (94) to thread through its accelerated reaction with benzylamine 105, forming amide rotaxane 144 (Fig. 47).222 Subsequent addition of a bulky N-Boc group forms 145, with the macrocycle shuttling to the oligo(ethylene glycol) collection thread. Substitution of the amide to form phenolic ester 146, followed by another crown ether threading active template reaction, generates [3]rotaxane 147.
Fig. 47 (a) A stepwise-operated molecular pump that makes use of successive metal-free active template transamidation reactions.222 Macrocycles are pumped onto the axles by a sequence of: (1) an active template transamidation reaction, (2) Boc-addition, and (3) substitution by an activated ester. (b) [5]Rotaxane 148 formed after two pumping cycles from a thread with pump motifs at either terminus. X-Ray crystal structure of 148. (c) [4]Rotaxane 149, in which the macrocycles have been threaded onto the axle in a controlled sequence. |
Every pumping cycle can add an additional macrocycle to the thread for each activated ester present, and so [5]rotaxane 148 is obtained after only two transamidation pumping cycles from a thread with pumping modules at both ends. In this high energy, kinetically locked, structure an array of weak CH–O hydrogen bonds between the components is apparent from X-ray crystallography (Fig. 47b). By carrying out pumping cycles with different crown ether derivatives, it was also possible to synthesise rotaxanes with specific sequences of threaded macrocycles, such as 149 (Fig. 47c).
Metal-free active template synthesis also enables the design of molecular pumps that operate autonomously, in which macrocycles are continuously pumped onto an axle, away from a state of equilibrium, powered by a chemical fuel.223 Catalysis-driven molecular pump 150 features a free benzylic amine at one end and a bulky stopper at the other (Fig. 48). The poor binding affinity of crown ether macrocycles for triazoles means that a vanishingly small proportion of macrocycles are threaded on the axle at equilibrium. However, an Fmoc-derived chemical fuel, 151, acts as an electrophile for a crown ether metal-free active template reaction, forming [2]rotaxane car-152 (the italicised prefix refers to the position of the ring on the axle). The Fmoc-carbamate group prevents macrocycle dethreading and the restricted steric environment around the carbamate site results in the macrocycle being displaced to the catchment region of the thread (tri-152). The presence of i-Pr2NH base in the reaction mixture triggers decomposition of the Fmoc-carbamate group faster when the bulky macrocycle is distant from the carbamate group. The trifluoromethyl group (shown in red in Fig. 48) is sufficiently large (a molecular ‘speed bump’) that it slows dethreading of the macrocycle from the catchment region off the open amine-terminated end of the thread.224 Therefore pumping of another crown ether onto the axle occurs faster than dethreading of the ring already on the axle, leading to the formation of [4]rotaxane 153. Macrocycles are thus continuously pumped onto the axle until all of the Fmoc-fuel is depleted, after which the rings dethread to regenerate the original ring equilibrium distribution in solution.
Fig. 48 An autonomous catalysis-driven artificial molecular pump 150 which operates continuously in the presence of chemical fuel 151.223 |
Goldup and co-workers found an unusual tandem process occurred during the active template CuAAC synthesis of [2]rotaxanes involving propargylic alcohol 154 as the alkyne substrate. When small bipyridine macrocycle 155 was used, a post-assembly rearrangement of the stabilised Cu(I)-triazolide interlocked derivative was triggered rendering the unexpected acrylamide rotaxane 156 (Fig. 49).225 Mechanistic studies suggest that mechanical bond stabilisation of the triazolide intermediate and the steric constraints imposed during rotaxane formation alters the chemoselectivity of the metal-catalysed reaction.
Fig. 49 Proposed mechanism of the tandem active template CuAAC rearrangement reported by Goldup and co-workers.225 |
Much of the early work in this field focussed on expanding the toolbox of catalytic reactions suitable for active template synthesis. A particularly attractive target remains finding a way to maintain the coordination of Pd(0) to a component throughout the catalytic cycle so that Pd(0)-mediated cross-coupling reactions can be used. Other challenges arise from functional group tolerance, and the ligand requirements of particular metal-catalysed reactions for active template synthesis if the preparation or handling of the building blocks is problematic.
Finally, while substantial progress has been made in the synthesis of interlocked molecules over the last four decades, practical applications of rotaxanes and catenanes remain scarce. The lack of requirements for permanent binding motifs may make active template synthesis more amenable for making practically useful interlocked structures than other synthetic strategies. Active template synthesis appears a promising route to rotaxanes and catenanes with applications in nano-electronics and biotechnology. Further exploration of different metal-mediated and metal-free active template methodologies, coupled with potential applications of newly accessible structure types, augurs well for the continued advancement18 of the chemistry of mechanically interlocked molecules.
This journal is © The Royal Society of Chemistry 2024 |