Introduction

The design of mechanically interlocked molecules (MIMs) has emerged as a cutting-edge research area over the past decades1,2, unveiling captivating breakthroughs in their potential applications, particularly highlighting their role as prototypes in artificial molecular machinery3,4,5,6,7,8,9. Since the first examples of catenanes10,11 and rotaxanes12 were reported, dependent on low-yielding “statistical methods”, substantial efforts have been made to develop innovative approaches to attain efficient syntheses of MIMs13,14,15,16,17,18,19,20,21,22,23. In the particular case of amide-based rotaxanes, whose assembly is guided by the establishment of hydrogen bonding interactions, noteworthy progress has been achieved in enhancing the efficiency of their preparation24. The use of future threads with a preorganized arrangement of hydrogen-bond accepting sites is essential to anneal around them benzylic amide-derived macrocycles. Thus, by following the well-known five-component clipping methodology for the assembly of this type of rotaxanes25, a variety of templates have been employed over the years, including succinamides26,27, small peptides28,29, bisnitrones30, squaraines31, organophosphorus species32, di(acylamino)pyrididines33,34,35, azodicarboxamides36,37, pyridyl-acyl hydrazones38,39, sulfur-containing motifs40,41, or glutaconamides42, with different level of efficiency (8-70% yield). It is remarkable the use of fumaramides as templates43,44, triggering the formation of the corresponding rotaxanes up to 97% isolated yield, the “world-record yield for amide-based rotaxanes” to date. In cases where the design of a controllable MIM requires the use of a low-yielding template, it is sometimes desirable to prepare an intermediate system that can be modified through a post-synthetic modification protocol after the interlocked architecture has been constructed45.

In this work, we present twofold vinylogous fumaramides, a class of conjugated bis(enaminones), as effective templates for assembling amide-based [2]rotaxanes (Fig. 1). The five-component clipping reaction of a series of bis(enaminone)-derived threads yielded the rotaxanes in exceptional yields. Comparative analysis with their fumaramide-containing partners in solution and solid state, alongside DFT calculations, confirmed the exceptional templating effect of the bis(enaminone) functionality. Additionally, we demonstrate its versatility through a double-stopper exchange45,46,47,48,49,50,51,52,53,54,55,56 and the direct synthesis of a three-station molecular shuttle, in which the position of the macrocycle along the thread was satisfyingly controlled by a chemical input.

Fig. 1: Conjugated bis-enaminones threads as templates for the formation of hydrogen-bonded rotaxanes and their synthetic post-modification via double stopper-exchange protocol.
figure 1

Proposed strategy for the building of rotaxanes using twofold vinylogous fumaramides and subsequent transformations through the addition and elimination of amines at the enaminone ends.

Full size image

Results and discussion

Use of conjugated bis(enaminone)-based threads for the formation of hydrogen-bonded rotaxanes

Initially, we computed the electrostatic potential maps for fumaramide thread 1a, which successfully templated the formation of the rotaxane 2a and other surrogates44,57,58, and its bis(enaminone)-based analogous 3a, plausible template for rotaxane 4a (Fig. 2). These analyses revealed negative regions around the oxygen atoms of both carbonyl groups, the hydrogen-bonding acceptor moieties. The corresponding maximum value was larger for thread 3a (−0.0844 au) than for thread 1a (−0.0674 au). Therefore, the electrostatic potential values for the isolated molecules predict the interaction energy of the tetraamide macrocycle with the bis(enaminone)-based thread 3a to be stronger than with 1a. With this promising data in hand, we experimentally tested a series of bis(enaminone) threads 3 as templates for the assembly of the corresponding rotaxanes 4.

Fig. 2: Comparison of electrostatic potential maps of threads 1a and 3a and their corresponding amide-based rotaxanes 2 and 4.
figure 2

Computed electrostatic potential map on the 0.001 au electron density isosurface of: a fumaramide-based thread 1a; b bis(enaminone)-based thread 3a. The values (au) of the MEP maxima (black dots) around the carbonyl group are indicated. Insets: fumaramide-based rotaxanes 2 (previously reported) and their bis(enaminone)-based analogous 4 (this work).

Full size image

Synthesis of threads 3 and rotaxanes 4

A series of threads 3 and their corresponding rotaxanes 4, differing only in the stoppers, were synthesized (Fig. 3). This selection of the stoppers aims to compare the templating ability of bis(enaminones) 3 with that of analogous fumaramides 143,44,59, also allowing the comparison of the internal dynamics and reactivity of bis(enaminone) rotaxanes 4 with that of the fumaramide rotaxanes 2. Starting from (E)-hex-3-ene-2,5-dione (5), its reaction with DMF·DMA yielded the conjugated bis(enaminone) 6, which is the precursor for the synthesis of threads 3 (see Supplementary Methods for further details). Nucleophilic substitution of the dimethylamino groups of 6 with the appropriate amine in refluxing EtOH delivered moderate yields of threads 3a-c (see Supplementary Notes 1 and 2). Finally, the five-component clipping reactions of the different threads 3 with p-xylylenediamine and isophthaloyl chloride, in the presence of Et3N, afforded rotaxanes 4 with high yields (see Supplementary Note 3). In all the cases, the presence of unreacted thread 3 was not observed (100% conversion), which shows their effectiveness as templates. In the case of the bis(enaminone)-based thread 3a, containing four benzyl groups as stoppers, an impressive yield of 90% of rotaxane 4a was isolated (with full conversion of the starting thread, see Supplementary Fig. S16), highly exceeding that of its fumaramide-based analog 2a (36% yield)44. Rotaxane 4b, however, was isolated in a 47% yield, whereas the fumaramide-based rotaxane 2b exhibited a 71% yield60. Having in mind the 100% conversion of thread 3b observed in 1H NMR of the reaction crude (Supplementary Fig. S22), the low isolated yield of rotaxane 4b was attributed to the kinetic instability of the system, due to the less bulky linear n-butyl groups, partially dethreading during the purification steps. Indeed, rotaxane 4b exhibited a complete disassembly in less than 3 min in DMSO-d6 at 100 °C (see Supplementary Note 10 and Supplementary Fig. S90), whereas rotaxane 2b was previously reported to undergo full dethreading nearly in an hour (k = 1.57 × 10−3 s−1 and t1/2 = 7.3 min)59, denoting the lower stability of the mechanical bond in 4b61. Rotaxane 4c was isolated in 95% yield (100% conversion of thread 2c, see Supplementary Fig. S28), a yield comparable to that of fumaramide-based 2c (97%)43. In solution, (1E,4E,7E) isomers of 3a,b and their rotaxanes 4a,b predominated, alongside traces of less stable isomers (1Z,4E,7E) and/or (1Z,4E,7Z). Conversely, thread 3c and rotaxane 4c, presenting a secondary enaminone, predominantly displayed (1Z,4E,7Z) isomers, most probably due to the intramolecular hydrogen bonds between the NH of the enaminones and the neighboring carbonyl oxygens62,63, conformation which does not influence the templating ability of thread 3c. Calculations also indicated a higher stability of (1Z,4E,7Z)-3c and (1Z,4E,7Z)-4c over their (1E,4E,7E) counterparts (see Supplementary Note 14, Supplementary Table S7).

Fig. 3: Synthesis of bis(enaminone)-based threads 3 and their corresponding hydrogen-bonded rotaxanes 4.
figure 3

Reaction conditions: (i) DMF·DMA, toluene, 110 °C, 12 h; (ii) amine, EtOH, 80 °C, 12 h; (iii) p-xylylenediamine (8 equiv), isophthaloyl dichloride (8 equiv), Et3N (24 equiv), CHCl3, 25 °C, 4 h. In brackets, conversion of the starting thread 3 measured by 1H NMR of the final reaction mixture.

Full size image

Single-crystal X-ray diffraction analysis and computational studies

The intercomponent interactions between the interlocked thread and the polyamide macrocycle in rotaxane 4a were deeply studied, both experimentally (solid state and solution) and computationally, and the results were compared with those obtained for rotaxane 2a. The structure of rotaxane 4a was elucidated by single-crystal X-ray diffraction (SCXRD) of suitable monocrystals grown by the slow diffusion of pentane into a CHCl3 solution of compound 4a (Fig. 4a, b and Supplementary Figs. S94–96). At the solid state, the thread in rotaxane 4a establishes two intramolecular hydrogen bonds with the macrocycle, triggering a distortion of the macrocyclic cavity, whereas the formation of two intermolecular hydrogen bonds between the macrocycles of neighboring molecules is also observed. In contrast, the macrocycle in tetrabenzylfumaramide rotaxane 2a establishes four bifurcated hydrogen bonds with the thread, adopting a highly symmetrical chair-like conformation64. Moreover, it is worth noting that the bis(enaminone) thread in 4a maintains a completely planar and highly stable (1E,4E,7E) conformation, as the one observed in solution, due to its extended π-π conjugation.

Fig. 4: X-Ray crystal structure of 4a and the intercomponent interaction energies for rotaxanes 2a and 4a.
figure 4

Structure of rotaxane 4a: a tilted view; b lateral view. Intramolecular hydrogen-bond lengths [Å] (and angles [°]): N3–H03…O1 1.99 (173°). For clarity, selected hydrogens atoms and solvent molecules (CHCl3) have been deleted. Computed non-covalent interaction energies for: c fumaramide-based rotaxane 2a; d bis(enaminone)-derived rotaxane 4a. Inset: BGR scale of surfaces according to values of sign(λ2)ρ (−0.07 to +0.07 au).

Full size image

In solution, notorious differences between rotaxanes 2a and 4a were also found. Variable-temperature 1H NMR experiments (see Supplementary Note 12) were carried out in non-competitive halogenated solvents (CDCl3 and CD2Cl2), and the energy barriers for the rotation of the macrocycle by using the coalescence method were calculated65,66,67. While the pirouetting motion of the macrocycle in rotaxane 2a was slow at a low-temperature regime, finding a coalescence temperature (Tc) of 323 K and a calculated energy barrier of 59.6 kJ·mol−1 (kc = 1537 s−1), in the case of the bis(enaminone)-based system 4a, the coalescence of the methylene protons of the macrocycle was observed at lower temperature (Tc = 238 K), with a calculated energy barrier of 45.1 kJ·mol−1 (kc = 633 s−1) (Supplementary Figs. S92, 93 and Supplementary Table S4). These data indicate the establishment of weaker interactions between the thread and the macrocycle in 4a than in 2a, although the yield obtained for 4a was much higher. In fact, the computed interaction energy between the macrocycle and the thread is larger for rotaxane 2a (−271.3 kJ mol−1) than for rotaxane 4a (−253.1 kJ mol−1). This difference is likely due to the ring-to-thread hydrogen-bonding pattern of both MIMs, as commented above, but also to the presence of two stabilizing aromatic-aromatic interactions between the benzyl groups at the thread and the isophthalamide groups at the macrocycle in rotaxane 2a, while rotaxane 4a shows only two weaker CH-π interactions, as revealed by the non-covalent interaction analyses of both rotaxanes (Fig. 4c, d)68. These interactions must be considered due to their significant influence on the internal dynamics of mechanically interlocked compounds in solution27,64,69.

Double stopper-exchange protocol

Having in mind the synthetic routes for threads 3, including the Me2NH displacement in compound 6 with different amines, we visualized a similar double stopper-exchange process in rotaxanes 4 that could leave undisturbed its mechanical bond. Rotaxane 4a was initially selected as the substrate for checking this protocol (Fig. 5). Fortunately, after exploring different solvents and amines, we successfully achieved the conversions of rotaxane 4a into various other interlocked derivatives with new stoppers (see Supplementary Note 4 for further details, Supplementary Tables S1, 2). The reaction of 4a with primary amines in tetrachloroethane as solvent gave the respective substituted rotaxane 4c (substituted with 2,2-diphenylethylamine) in a 95% yield and 4d (substituted with 9-(aminomethyl)anthracene) in 60% yield (Fig. 5a). In this latter case, the reaction was not complete, recovering 30% of the remained starting material 4a. In contrast, no rotaxanes were obtained when the reaction was performed with a secondary amine, only isolating the respective newly formed thread (Supplementary Table S3) and observing the appearance of a white solid. This solid corresponded to the highly insoluble free tetraamide macrocycle70, which was filtered, washed, and stored for potential future applications71. In contrast, no rotaxanes were obtained when the reaction was performed with a secondary amine, only isolating the respective newly formed thread (Supplementary Table S3). Moreover, rotaxane 4c was unreactive in the presence of a wide range of amines (see Supplementary Note 5), precluding further chemical modification by a stopper-exchange methodology. To highlight the potential of this protocol, we straightforwardly obtained the three-station molecular shuttle 5 in 70% yield, by reaction of rotaxane 4a with N1-(4-(aminomethyl)benzyl)-N4,N4-dibenzylsuccinamide (Fig. 5b) (see Supplementary Note 6). It is important to highlight that the obtention of 5 by following a five-component clipping methodology from its thread 6 (see Supplementary Note 7 for synthetic details) was totally unselective, obtaining a mixture of interlocked species ([2]rotaxane 5, in 13% yield, together with the corresponding [3]rotaxane in 58%, and some amount of the [4]rotaxane), highly challenging to separate (see Supplementary Note 8 for further details). The lack of selectivity of the clipping method in controlling the order of the rotaxane highly enhances the importance of the double stopper-exchange protocol for the obtention of interlocked species.

Fig. 5: Double stopper-exchange protocol for the synthesis of rotaxanes 4c-d and shuttle 5, including shuttling via DA and retroDA reactions.
figure 5

a Synthesis of rotaxanes 4c-d; b Synthesis of shuttle 5 and its chemical interconversion. Reaction conditions: (i) 2,2-diphenylethylamine (2.2 equiv), TCE, 100 °C, 12 h; (ii) 9-(aminomethyl)anthracene (2.2 equiv), TCE, 100 °C, 12 h; (iii) N1-(4-(aminomethyl)benzyl)-N4,N4-dibenzylsuccinamide (3 equiv), THF, 60 °C, 5 h (70%); (iv) p-xylylenediamine (8 equiv), isophthaloyl dichloride (8 equiv), Et3N (24 equiv), CHCl3, 25 °C, 4 h (13%); (v) cyclopentadiene, DMSO, 70 °C, 12 h (39%); (vi) 165 °C, 0.5 mm Hg, 1 h (65%).

Full size image

Chemically induced reversible control of macrocycle location in rotaxane 5

The comparison of the 1H NMR spectra of shuttle 5 and its free thread 6 allows us to confirm that the macrocycle is located over the central station (signal Hg is highly shifted to lower chemical shift, Δδ = −1.13 ppm) (Fig. 6a, b, see Fig. 4 for lettering). We estimated the occupancy of this station in rotaxane 5 by comparing the difference of chemical shift of the central double bond in thread 3c and its rotaxane 4c (Δδ = −1.36 ppm, 100% occupancy), being 83%. We then focused on the positional control of the macrocycle in rotaxane 572,73,74. The presence of the central olefin at the bis(enaminone) function in 5 (in green color) first led us to think of a photochemical stimulus. While fumaramide-based rotaxanes 2 isomerized to the corresponding maleamide derivatives under UV irradiation75, rotaxane 4a, selected as model compound, remained unreactive after irradiating under 254 nm or 320 nm light (see Supplementary Note 11, Supplementary Fig. S91). As an alternative, a Diels–Alder (DA) reaction was satisfyingly performed, where the central C = C bond (signal Hg) in 5 reacted with cyclopentadiene forming a 1:1 diasteroisomeric mixture of cycloadduct Cp-5 (Fig. 5b)76,77. Analysis of its 1H NMR spectra revealed that the macrocycle is now displaced towards one of the succinamide stations (Ha+b, orange color). The DA adduct (in red color, Greek lettering) is bulky enough to act as a steric barrier, compartmentalizing the system and precluding the macrocycle to freely move between the two succinamides, keeping one of these occupied, and the second one unoccupied, as shown in Fig. 6. As expected, signals related to the DA cycloadduct appeared at similar chemical shifts at both thread Cp-6 (not shown in Fig. 4) and shuttle Cp-5 (Greek lettering, Fig. 6c, d). Remarkably, the retro-Diels–Alder reaction of Cp-5 under thermal and high vacuum conditions, resulted in the recovery of the starting rotaxane 5 (see Supplementary Note 9 for further synthetic details). Thus, we successfully achieved almost full control over the location of the macrocycle in shuttle 5, by performing a reversible chemical reaction over the interlocked conjugated bis(enaminone) thread.

Fig. 6: NMR analysis for the assessment of ring shuttling in rotaxane 5.
figure 6

Stacked 1H NMR (400 MHz, CDCl3, 298 K) of: a thread 6; b shuttle 5; c thread Cp-6 (see structure at the SI); d shuttle Cp-5. Signals related to the Diels–Alder adduct are colored in red. Signals related to the conjugated bis(enaminone) station are colored in green. Signals related to the succinamide stations are colored in orange. Signals related to the tetraamide macrocycle are colored in light blue. For lettering, see Fig. 5b.

Full size image

Conclusions

In this work, we have reported the use of the conjugated bis(enaminone) motif as an efficient template for the formation of hydrogen-bonded rotaxanes in excellent yields (up to 100% conversion of the starting thread). The preorganized spatial arrangement of the hydrogen-bonded acceptors and the electronic features of this rigid binding site, like those of the fumaramide-based template, are crucial for this exceptional behavior, as demonstrated both experimentally and computationally. We have also developed an appealing chemical route based on a double stopper-exchange protocol that allowed to form new interlocked species while keeping untouched the mechanical bond. By following this protocol, a molecular shuttle was selectively obtained, circumventing the absence of selectivity of the clipping of a multi-station linear component. The control of the translational motion of the macrocycle was reversibly tamed by using a chemical stimulus, a DA/retroDA reaction occurring at the central core of the thread. Thus, these templates are promising candidates for the future design of more sophisticated mechanically interlocked molecules, opening exciting avenues in the realm of functional molecular architectures.

Methods

General experimental section (Supplementary Note 1)

Unless stated otherwise, all reagents were purchased from Aldrich Chemicals and used without further purification. HPLC grade solvents (Scharlab) were nitrogen saturated and were dried and deoxygenated using an Innovative Technology Inc. Pure-Solv 400 Solvent Purification System. Column chromatography was carried out using silica gel (60 Å, 70–200 μm, SDS) or aluminum oxide (activated, neutral) as stationary phase, and TLC was performed on precoated silica gel on aluminum cards (0.25 mm thick, with fluorescent indicator 254 nm, Fluka) and observed under UV light. All melting points were determined on a Kofler hot-plate melting point apparatus and are uncorrected. 1H and 13C-NMR spectra were recorded on Bruker Advance 300 and 400 MHz instruments. 1H NMR chemical shifts are reported relative to Me4Si and were referenced via residual proton resonances of the corresponding deuterated solvent, whereas 13C-NMR spectra are reported relative to Me4Si using the carbon signals of the deuterated solvent. Signals in the 1H and 13C NMR spectra of the synthesized compounds were assigned with the aid of DEPT, APT and COSY. Abbreviations of coupling patterns are as follows: br, broad; s, singlet; d, doublet; t, triplet; q, quadruplet; m, multiplet. The deuterated solvent CDCl3 was dried over CaH2 and stored with molecular sieves prior to use. Coupling constants (J) are expressed in Hz. High-resolution mass spectra (HRMS) were obtained using a time-of-flight (TOF) instrument equipped with electrospray ionization (ESI). Single-crystal X-ray structures were collected in Bruker D8QUEST diffractometer.

Protocol for the double stopper-exchange of rotaxane 4a

A solution of rotaxane 4a (1 equiv.) and selected amine (2.2 equiv.) in a selected solvent was stirred for 12 h at 100 °C. After this time, the solvent was removed, and the resulting solid crude was subjected to column chromatography on silica gel using a CHCl3/MeOH mixture as eluent to give the title product (See Supplementary Note 4).

Diels-Alder and retro-DA reactions of thread 6 and rotaxane 5 with cyclopentadiene

A solution of the corresponding thread 6 or rotaxane 5 (1 eq) and freshly cracked and distilled cyclopentadiene (10 equiv.) in DMSO under N2 atmosphere was heated at 70 °C for 12 h. After this time, the reaction mixture was extracted with CHCl3 and washed with a saturated NaCl solution. The resulting residue was purified by column chromatography using a mixture of CHCl3/MeOH 95/5 (v/v) (See Supplementary Note 9).

Computational methods

Geometries of the molecules were optimized by using the wB97X-D hybrid-functional with the cc-PVDZ basis set. The nature of minimum and transition structures of all stationary points on the potential energy surface was confirmed by frequency analysis at the same level of theory. The stability of the resulting wave functions was checked for all the optimized structures. Solvent effects were calculated with the PCM solvation model with chloroform parameters. The ultrafine grid implemented in Gaussian 16 C. 01 was used. The counterpoise method of Boys and Bernardi was employed to minimize the error of overlapping bases (BSSE) and was used to compute the corrected interaction energies between thread and macrocycle at the [2]rotaxanes. Non-covalent interaction analysis was performed using the NCI plot program (version 4.2.1) (See Supplementary Notes 13–15).

Full details of experimental and theoretical procedures are included in the Supplementary Information. Deposition numbers CCDC 2361005 (for 4a) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.